Upload folder using huggingface_hub
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- mlc-chat-config.json +90 -0
- ndarray-cache.json +2726 -0
- params_shard_0.bin +3 -0
- params_shard_1.bin +3 -0
- params_shard_10.bin +3 -0
- params_shard_100.bin +3 -0
- params_shard_101.bin +3 -0
- params_shard_102.bin +3 -0
- params_shard_103.bin +3 -0
- params_shard_104.bin +3 -0
- params_shard_105.bin +3 -0
- params_shard_106.bin +3 -0
- params_shard_107.bin +3 -0
- params_shard_108.bin +3 -0
- params_shard_109.bin +3 -0
- params_shard_11.bin +3 -0
- params_shard_110.bin +3 -0
- params_shard_111.bin +3 -0
- params_shard_112.bin +3 -0
- params_shard_12.bin +3 -0
- params_shard_13.bin +3 -0
- params_shard_14.bin +3 -0
- params_shard_15.bin +3 -0
- params_shard_16.bin +3 -0
- params_shard_17.bin +3 -0
- params_shard_18.bin +3 -0
- params_shard_19.bin +3 -0
- params_shard_2.bin +3 -0
- params_shard_20.bin +3 -0
- params_shard_21.bin +3 -0
- params_shard_22.bin +3 -0
- params_shard_23.bin +3 -0
- params_shard_24.bin +3 -0
- params_shard_25.bin +3 -0
- params_shard_26.bin +3 -0
- params_shard_27.bin +3 -0
- params_shard_28.bin +3 -0
- params_shard_29.bin +3 -0
- params_shard_3.bin +3 -0
- params_shard_30.bin +3 -0
- params_shard_31.bin +3 -0
- params_shard_32.bin +3 -0
- params_shard_33.bin +3 -0
- params_shard_34.bin +3 -0
- params_shard_35.bin +3 -0
- params_shard_36.bin +3 -0
- params_shard_37.bin +3 -0
- params_shard_38.bin +3 -0
- params_shard_39.bin +3 -0
- params_shard_4.bin +3 -0
mlc-chat-config.json
ADDED
|
@@ -0,0 +1,90 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"version": "0.1.0",
|
| 3 |
+
"model_type": "llama",
|
| 4 |
+
"quantization": "q0f16",
|
| 5 |
+
"model_config": {
|
| 6 |
+
"hidden_size": 3072,
|
| 7 |
+
"intermediate_size": 8192,
|
| 8 |
+
"num_attention_heads": 24,
|
| 9 |
+
"num_hidden_layers": 28,
|
| 10 |
+
"rms_norm_eps": 1e-05,
|
| 11 |
+
"vocab_size": 128256,
|
| 12 |
+
"tie_word_embeddings": true,
|
| 13 |
+
"position_embedding_base": 500000.0,
|
| 14 |
+
"rope_scaling": {
|
| 15 |
+
"factor": 32.0,
|
| 16 |
+
"high_freq_factor": 4.0,
|
| 17 |
+
"low_freq_factor": 1.0,
|
| 18 |
+
"original_max_position_embeddings": 8192,
|
| 19 |
+
"rope_type": "llama3"
|
| 20 |
+
},
|
| 21 |
+
"context_window_size": 131072,
|
| 22 |
+
"prefill_chunk_size": 8192,
|
| 23 |
+
"num_key_value_heads": 8,
|
| 24 |
+
"head_dim": 128,
|
| 25 |
+
"tensor_parallel_shards": 1,
|
| 26 |
+
"pipeline_parallel_stages": 1,
|
| 27 |
+
"max_batch_size": 128
|
| 28 |
+
},
|
| 29 |
+
"vocab_size": 128256,
|
| 30 |
+
"context_window_size": 131072,
|
| 31 |
+
"sliding_window_size": -1,
|
| 32 |
+
"prefill_chunk_size": 8192,
|
| 33 |
+
"attention_sink_size": -1,
|
| 34 |
+
"tensor_parallel_shards": 1,
|
| 35 |
+
"pipeline_parallel_stages": 1,
|
| 36 |
+
"temperature": 0.6,
|
| 37 |
+
"presence_penalty": 0.0,
|
| 38 |
+
"frequency_penalty": 0.0,
|
| 39 |
+
"repetition_penalty": 1.0,
|
| 40 |
+
"top_p": 0.9,
|
| 41 |
+
"tokenizer_files": [
|
| 42 |
+
"tokenizer.json",
|
| 43 |
+
"tokenizer_config.json"
|
| 44 |
+
],
|
| 45 |
+
"tokenizer_info": {
|
| 46 |
+
"token_postproc_method": "byte_level",
|
| 47 |
+
"prepend_space_in_encode": false,
|
| 48 |
+
"strip_space_in_decode": false
|
| 49 |
+
},
|
| 50 |
+
"conv_template": {
|
| 51 |
+
"name": "llama-3_1",
|
| 52 |
+
"system_template": "<|start_header_id|>system<|end_header_id|>\n\n{system_message}<|eot_id|>",
|
| 53 |
+
"system_message": "You are a helpful, respectful and honest assistant.",
|
| 54 |
+
"system_prefix_token_ids": [
|
| 55 |
+
128000
|
| 56 |
+
],
|
| 57 |
+
"add_role_after_system_message": true,
|
| 58 |
+
"roles": {
|
| 59 |
+
"user": "<|start_header_id|>user",
|
| 60 |
+
"assistant": "<|start_header_id|>assistant",
|
| 61 |
+
"tool": "<|start_header_id|>ipython"
|
| 62 |
+
},
|
| 63 |
+
"role_templates": {
|
| 64 |
+
"user": "{user_message}",
|
| 65 |
+
"assistant": "{assistant_message}",
|
| 66 |
+
"tool": "{tool_message}"
|
| 67 |
+
},
|
| 68 |
+
"messages": [],
|
| 69 |
+
"seps": [
|
| 70 |
+
"<|eot_id|>"
|
| 71 |
+
],
|
| 72 |
+
"role_content_sep": "<|end_header_id|>\n\n",
|
| 73 |
+
"role_empty_sep": "<|end_header_id|>\n\n",
|
| 74 |
+
"stop_str": [],
|
| 75 |
+
"stop_token_ids": [
|
| 76 |
+
128001,
|
| 77 |
+
128008,
|
| 78 |
+
128009
|
| 79 |
+
],
|
| 80 |
+
"function_string": "",
|
| 81 |
+
"use_function_calling": false
|
| 82 |
+
},
|
| 83 |
+
"pad_token_id": 0,
|
| 84 |
+
"bos_token_id": 128000,
|
| 85 |
+
"eos_token_id": [
|
| 86 |
+
128001,
|
| 87 |
+
128008,
|
| 88 |
+
128009
|
| 89 |
+
]
|
| 90 |
+
}
|
ndarray-cache.json
ADDED
|
@@ -0,0 +1,2726 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"metadata": {
|
| 3 |
+
"ParamSize": 170,
|
| 4 |
+
"ParamBytes": 6425499648.0,
|
| 5 |
+
"BitsPerParam": 16.0
|
| 6 |
+
},
|
| 7 |
+
"records": [
|
| 8 |
+
{
|
| 9 |
+
"dataPath": "params_shard_0.bin",
|
| 10 |
+
"format": "raw-shard",
|
| 11 |
+
"nbytes": 788004864,
|
| 12 |
+
"records": [
|
| 13 |
+
{
|
| 14 |
+
"name": "model.embed_tokens.weight",
|
| 15 |
+
"shape": [
|
| 16 |
+
128256,
|
| 17 |
+
3072
|
| 18 |
+
],
|
| 19 |
+
"dtype": "float16",
|
| 20 |
+
"format": "f32-to-bf16",
|
| 21 |
+
"nbytes": 788004864,
|
| 22 |
+
"byteOffset": 0
|
| 23 |
+
}
|
| 24 |
+
],
|
| 25 |
+
"md5sum": "ad5b241e75194e932113ce287edfd3e5"
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"dataPath": "params_shard_1.bin",
|
| 29 |
+
"format": "raw-shard",
|
| 30 |
+
"nbytes": 50331648,
|
| 31 |
+
"records": [
|
| 32 |
+
{
|
| 33 |
+
"name": "model.layers.0.mlp.down_proj.weight",
|
| 34 |
+
"shape": [
|
| 35 |
+
3072,
|
| 36 |
+
8192
|
| 37 |
+
],
|
| 38 |
+
"dtype": "float16",
|
| 39 |
+
"format": "f32-to-bf16",
|
| 40 |
+
"nbytes": 50331648,
|
| 41 |
+
"byteOffset": 0
|
| 42 |
+
}
|
| 43 |
+
],
|
| 44 |
+
"md5sum": "49b4d4c409da6bdd739e9645457fbc28"
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"dataPath": "params_shard_2.bin",
|
| 48 |
+
"format": "raw-shard",
|
| 49 |
+
"nbytes": 100663296,
|
| 50 |
+
"records": [
|
| 51 |
+
{
|
| 52 |
+
"name": "model.layers.0.mlp.gate_up_proj.weight",
|
| 53 |
+
"shape": [
|
| 54 |
+
16384,
|
| 55 |
+
3072
|
| 56 |
+
],
|
| 57 |
+
"dtype": "float16",
|
| 58 |
+
"format": "f32-to-bf16",
|
| 59 |
+
"nbytes": 100663296,
|
| 60 |
+
"byteOffset": 0
|
| 61 |
+
}
|
| 62 |
+
],
|
| 63 |
+
"md5sum": "08f668bf0839e1a1a1e49e8b4d31d8a0"
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"dataPath": "params_shard_3.bin",
|
| 67 |
+
"format": "raw-shard",
|
| 68 |
+
"nbytes": 18874368,
|
| 69 |
+
"records": [
|
| 70 |
+
{
|
| 71 |
+
"name": "model.layers.0.self_attn.o_proj.weight",
|
| 72 |
+
"shape": [
|
| 73 |
+
3072,
|
| 74 |
+
3072
|
| 75 |
+
],
|
| 76 |
+
"dtype": "float16",
|
| 77 |
+
"format": "f32-to-bf16",
|
| 78 |
+
"nbytes": 18874368,
|
| 79 |
+
"byteOffset": 0
|
| 80 |
+
}
|
| 81 |
+
],
|
| 82 |
+
"md5sum": "6a833ac91147731d0dc694797387c655"
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"dataPath": "params_shard_4.bin",
|
| 86 |
+
"format": "raw-shard",
|
| 87 |
+
"nbytes": 50331648,
|
| 88 |
+
"records": [
|
| 89 |
+
{
|
| 90 |
+
"name": "model.layers.1.mlp.down_proj.weight",
|
| 91 |
+
"shape": [
|
| 92 |
+
3072,
|
| 93 |
+
8192
|
| 94 |
+
],
|
| 95 |
+
"dtype": "float16",
|
| 96 |
+
"format": "f32-to-bf16",
|
| 97 |
+
"nbytes": 50331648,
|
| 98 |
+
"byteOffset": 0
|
| 99 |
+
}
|
| 100 |
+
],
|
| 101 |
+
"md5sum": "223470f4bf8228bb0407825886d53c4e"
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"dataPath": "params_shard_5.bin",
|
| 105 |
+
"format": "raw-shard",
|
| 106 |
+
"nbytes": 100663296,
|
| 107 |
+
"records": [
|
| 108 |
+
{
|
| 109 |
+
"name": "model.layers.1.mlp.gate_up_proj.weight",
|
| 110 |
+
"shape": [
|
| 111 |
+
16384,
|
| 112 |
+
3072
|
| 113 |
+
],
|
| 114 |
+
"dtype": "float16",
|
| 115 |
+
"format": "f32-to-bf16",
|
| 116 |
+
"nbytes": 100663296,
|
| 117 |
+
"byteOffset": 0
|
| 118 |
+
}
|
| 119 |
+
],
|
| 120 |
+
"md5sum": "c3c4fc75c26d9e1c28aa943caaffae43"
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"dataPath": "params_shard_6.bin",
|
| 124 |
+
"format": "raw-shard",
|
| 125 |
+
"nbytes": 31457280,
|
| 126 |
+
"records": [
|
| 127 |
+
{
|
| 128 |
+
"name": "model.layers.1.self_attn.qkv_proj.weight",
|
| 129 |
+
"shape": [
|
| 130 |
+
5120,
|
| 131 |
+
3072
|
| 132 |
+
],
|
| 133 |
+
"dtype": "float16",
|
| 134 |
+
"format": "f32-to-bf16",
|
| 135 |
+
"nbytes": 31457280,
|
| 136 |
+
"byteOffset": 0
|
| 137 |
+
}
|
| 138 |
+
],
|
| 139 |
+
"md5sum": "ed6d793fc2c6b84e239a989ccf0f0c63"
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"dataPath": "params_shard_7.bin",
|
| 143 |
+
"format": "raw-shard",
|
| 144 |
+
"nbytes": 18874368,
|
| 145 |
+
"records": [
|
| 146 |
+
{
|
| 147 |
+
"name": "model.layers.1.self_attn.o_proj.weight",
|
| 148 |
+
"shape": [
|
| 149 |
+
3072,
|
| 150 |
+
3072
|
| 151 |
+
],
|
| 152 |
+
"dtype": "float16",
|
| 153 |
+
"format": "f32-to-bf16",
|
| 154 |
+
"nbytes": 18874368,
|
| 155 |
+
"byteOffset": 0
|
| 156 |
+
}
|
| 157 |
+
],
|
| 158 |
+
"md5sum": "0206bc3ab17cdfdc7ce7c1f5eec31f11"
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"dataPath": "params_shard_8.bin",
|
| 162 |
+
"format": "raw-shard",
|
| 163 |
+
"nbytes": 50331648,
|
| 164 |
+
"records": [
|
| 165 |
+
{
|
| 166 |
+
"name": "model.layers.10.mlp.down_proj.weight",
|
| 167 |
+
"shape": [
|
| 168 |
+
3072,
|
| 169 |
+
8192
|
| 170 |
+
],
|
| 171 |
+
"dtype": "float16",
|
| 172 |
+
"format": "f32-to-bf16",
|
| 173 |
+
"nbytes": 50331648,
|
| 174 |
+
"byteOffset": 0
|
| 175 |
+
}
|
| 176 |
+
],
|
| 177 |
+
"md5sum": "e54a86424568408412013b4061c52457"
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"dataPath": "params_shard_9.bin",
|
| 181 |
+
"format": "raw-shard",
|
| 182 |
+
"nbytes": 100663296,
|
| 183 |
+
"records": [
|
| 184 |
+
{
|
| 185 |
+
"name": "model.layers.10.mlp.gate_up_proj.weight",
|
| 186 |
+
"shape": [
|
| 187 |
+
16384,
|
| 188 |
+
3072
|
| 189 |
+
],
|
| 190 |
+
"dtype": "float16",
|
| 191 |
+
"format": "f32-to-bf16",
|
| 192 |
+
"nbytes": 100663296,
|
| 193 |
+
"byteOffset": 0
|
| 194 |
+
}
|
| 195 |
+
],
|
| 196 |
+
"md5sum": "2e562be0f2ef06acca5d5dc2d4d69083"
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"dataPath": "params_shard_10.bin",
|
| 200 |
+
"format": "raw-shard",
|
| 201 |
+
"nbytes": 31457280,
|
| 202 |
+
"records": [
|
| 203 |
+
{
|
| 204 |
+
"name": "model.layers.10.self_attn.qkv_proj.weight",
|
| 205 |
+
"shape": [
|
| 206 |
+
5120,
|
| 207 |
+
3072
|
| 208 |
+
],
|
| 209 |
+
"dtype": "float16",
|
| 210 |
+
"format": "f32-to-bf16",
|
| 211 |
+
"nbytes": 31457280,
|
| 212 |
+
"byteOffset": 0
|
| 213 |
+
}
|
| 214 |
+
],
|
| 215 |
+
"md5sum": "345e4cf1097065e91b9917fc79267ad9"
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"dataPath": "params_shard_11.bin",
|
| 219 |
+
"format": "raw-shard",
|
| 220 |
+
"nbytes": 18874368,
|
| 221 |
+
"records": [
|
| 222 |
+
{
|
| 223 |
+
"name": "model.layers.10.self_attn.o_proj.weight",
|
| 224 |
+
"shape": [
|
| 225 |
+
3072,
|
| 226 |
+
3072
|
| 227 |
+
],
|
| 228 |
+
"dtype": "float16",
|
| 229 |
+
"format": "f32-to-bf16",
|
| 230 |
+
"nbytes": 18874368,
|
| 231 |
+
"byteOffset": 0
|
| 232 |
+
}
|
| 233 |
+
],
|
| 234 |
+
"md5sum": "81bfc646c8625007af2981512c5367a8"
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"dataPath": "params_shard_12.bin",
|
| 238 |
+
"format": "raw-shard",
|
| 239 |
+
"nbytes": 50331648,
|
| 240 |
+
"records": [
|
| 241 |
+
{
|
| 242 |
+
"name": "model.layers.11.mlp.down_proj.weight",
|
| 243 |
+
"shape": [
|
| 244 |
+
3072,
|
| 245 |
+
8192
|
| 246 |
+
],
|
| 247 |
+
"dtype": "float16",
|
| 248 |
+
"format": "f32-to-bf16",
|
| 249 |
+
"nbytes": 50331648,
|
| 250 |
+
"byteOffset": 0
|
| 251 |
+
}
|
| 252 |
+
],
|
| 253 |
+
"md5sum": "a2b2fbdcc692dea5cabb310d9c6169ef"
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"dataPath": "params_shard_13.bin",
|
| 257 |
+
"format": "raw-shard",
|
| 258 |
+
"nbytes": 100663296,
|
| 259 |
+
"records": [
|
| 260 |
+
{
|
| 261 |
+
"name": "model.layers.11.mlp.gate_up_proj.weight",
|
| 262 |
+
"shape": [
|
| 263 |
+
16384,
|
| 264 |
+
3072
|
| 265 |
+
],
|
| 266 |
+
"dtype": "float16",
|
| 267 |
+
"format": "f32-to-bf16",
|
| 268 |
+
"nbytes": 100663296,
|
| 269 |
+
"byteOffset": 0
|
| 270 |
+
}
|
| 271 |
+
],
|
| 272 |
+
"md5sum": "2fb0350bef3892ee6cf7d17d53d512ee"
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"dataPath": "params_shard_14.bin",
|
| 276 |
+
"format": "raw-shard",
|
| 277 |
+
"nbytes": 31457280,
|
| 278 |
+
"records": [
|
| 279 |
+
{
|
| 280 |
+
"name": "model.layers.11.self_attn.qkv_proj.weight",
|
| 281 |
+
"shape": [
|
| 282 |
+
5120,
|
| 283 |
+
3072
|
| 284 |
+
],
|
| 285 |
+
"dtype": "float16",
|
| 286 |
+
"format": "f32-to-bf16",
|
| 287 |
+
"nbytes": 31457280,
|
| 288 |
+
"byteOffset": 0
|
| 289 |
+
}
|
| 290 |
+
],
|
| 291 |
+
"md5sum": "287c107bf9ed9301cc71d715204f9be3"
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"dataPath": "params_shard_15.bin",
|
| 295 |
+
"format": "raw-shard",
|
| 296 |
+
"nbytes": 18874368,
|
| 297 |
+
"records": [
|
| 298 |
+
{
|
| 299 |
+
"name": "model.layers.11.self_attn.o_proj.weight",
|
| 300 |
+
"shape": [
|
| 301 |
+
3072,
|
| 302 |
+
3072
|
| 303 |
+
],
|
| 304 |
+
"dtype": "float16",
|
| 305 |
+
"format": "f32-to-bf16",
|
| 306 |
+
"nbytes": 18874368,
|
| 307 |
+
"byteOffset": 0
|
| 308 |
+
}
|
| 309 |
+
],
|
| 310 |
+
"md5sum": "91507ab82a0d6dc77d41924a1d9b8863"
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"dataPath": "params_shard_16.bin",
|
| 314 |
+
"format": "raw-shard",
|
| 315 |
+
"nbytes": 50331648,
|
| 316 |
+
"records": [
|
| 317 |
+
{
|
| 318 |
+
"name": "model.layers.12.mlp.down_proj.weight",
|
| 319 |
+
"shape": [
|
| 320 |
+
3072,
|
| 321 |
+
8192
|
| 322 |
+
],
|
| 323 |
+
"dtype": "float16",
|
| 324 |
+
"format": "f32-to-bf16",
|
| 325 |
+
"nbytes": 50331648,
|
| 326 |
+
"byteOffset": 0
|
| 327 |
+
}
|
| 328 |
+
],
|
| 329 |
+
"md5sum": "22e107f1422969c7ae300ec3429a19ab"
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"dataPath": "params_shard_17.bin",
|
| 333 |
+
"format": "raw-shard",
|
| 334 |
+
"nbytes": 100663296,
|
| 335 |
+
"records": [
|
| 336 |
+
{
|
| 337 |
+
"name": "model.layers.12.mlp.gate_up_proj.weight",
|
| 338 |
+
"shape": [
|
| 339 |
+
16384,
|
| 340 |
+
3072
|
| 341 |
+
],
|
| 342 |
+
"dtype": "float16",
|
| 343 |
+
"format": "f32-to-bf16",
|
| 344 |
+
"nbytes": 100663296,
|
| 345 |
+
"byteOffset": 0
|
| 346 |
+
}
|
| 347 |
+
],
|
| 348 |
+
"md5sum": "5d1098cdba5e4706a47b7d5dd9a73520"
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"dataPath": "params_shard_18.bin",
|
| 352 |
+
"format": "raw-shard",
|
| 353 |
+
"nbytes": 31457280,
|
| 354 |
+
"records": [
|
| 355 |
+
{
|
| 356 |
+
"name": "model.layers.12.self_attn.qkv_proj.weight",
|
| 357 |
+
"shape": [
|
| 358 |
+
5120,
|
| 359 |
+
3072
|
| 360 |
+
],
|
| 361 |
+
"dtype": "float16",
|
| 362 |
+
"format": "f32-to-bf16",
|
| 363 |
+
"nbytes": 31457280,
|
| 364 |
+
"byteOffset": 0
|
| 365 |
+
}
|
| 366 |
+
],
|
| 367 |
+
"md5sum": "2cc80b6c74f4edc400f1c5d56ac19925"
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"dataPath": "params_shard_19.bin",
|
| 371 |
+
"format": "raw-shard",
|
| 372 |
+
"nbytes": 18874368,
|
| 373 |
+
"records": [
|
| 374 |
+
{
|
| 375 |
+
"name": "model.layers.12.self_attn.o_proj.weight",
|
| 376 |
+
"shape": [
|
| 377 |
+
3072,
|
| 378 |
+
3072
|
| 379 |
+
],
|
| 380 |
+
"dtype": "float16",
|
| 381 |
+
"format": "f32-to-bf16",
|
| 382 |
+
"nbytes": 18874368,
|
| 383 |
+
"byteOffset": 0
|
| 384 |
+
}
|
| 385 |
+
],
|
| 386 |
+
"md5sum": "2a21ff1977220b6c5f63047b288e54e3"
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"dataPath": "params_shard_20.bin",
|
| 390 |
+
"format": "raw-shard",
|
| 391 |
+
"nbytes": 50331648,
|
| 392 |
+
"records": [
|
| 393 |
+
{
|
| 394 |
+
"name": "model.layers.13.mlp.down_proj.weight",
|
| 395 |
+
"shape": [
|
| 396 |
+
3072,
|
| 397 |
+
8192
|
| 398 |
+
],
|
| 399 |
+
"dtype": "float16",
|
| 400 |
+
"format": "f32-to-bf16",
|
| 401 |
+
"nbytes": 50331648,
|
| 402 |
+
"byteOffset": 0
|
| 403 |
+
}
|
| 404 |
+
],
|
| 405 |
+
"md5sum": "e2713d579e29395c7252668f13274808"
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"dataPath": "params_shard_21.bin",
|
| 409 |
+
"format": "raw-shard",
|
| 410 |
+
"nbytes": 100663296,
|
| 411 |
+
"records": [
|
| 412 |
+
{
|
| 413 |
+
"name": "model.layers.13.mlp.gate_up_proj.weight",
|
| 414 |
+
"shape": [
|
| 415 |
+
16384,
|
| 416 |
+
3072
|
| 417 |
+
],
|
| 418 |
+
"dtype": "float16",
|
| 419 |
+
"format": "f32-to-bf16",
|
| 420 |
+
"nbytes": 100663296,
|
| 421 |
+
"byteOffset": 0
|
| 422 |
+
}
|
| 423 |
+
],
|
| 424 |
+
"md5sum": "7ca0318836b0573b059dc4091244de0f"
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"dataPath": "params_shard_22.bin",
|
| 428 |
+
"format": "raw-shard",
|
| 429 |
+
"nbytes": 31457280,
|
| 430 |
+
"records": [
|
| 431 |
+
{
|
| 432 |
+
"name": "model.layers.13.self_attn.qkv_proj.weight",
|
| 433 |
+
"shape": [
|
| 434 |
+
5120,
|
| 435 |
+
3072
|
| 436 |
+
],
|
| 437 |
+
"dtype": "float16",
|
| 438 |
+
"format": "f32-to-bf16",
|
| 439 |
+
"nbytes": 31457280,
|
| 440 |
+
"byteOffset": 0
|
| 441 |
+
}
|
| 442 |
+
],
|
| 443 |
+
"md5sum": "6a6ab8255ffa0f2d997070e1ace2a330"
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"dataPath": "params_shard_23.bin",
|
| 447 |
+
"format": "raw-shard",
|
| 448 |
+
"nbytes": 18874368,
|
| 449 |
+
"records": [
|
| 450 |
+
{
|
| 451 |
+
"name": "model.layers.13.self_attn.o_proj.weight",
|
| 452 |
+
"shape": [
|
| 453 |
+
3072,
|
| 454 |
+
3072
|
| 455 |
+
],
|
| 456 |
+
"dtype": "float16",
|
| 457 |
+
"format": "f32-to-bf16",
|
| 458 |
+
"nbytes": 18874368,
|
| 459 |
+
"byteOffset": 0
|
| 460 |
+
}
|
| 461 |
+
],
|
| 462 |
+
"md5sum": "755ef4e485bf7e00cf5f19fa9171fc13"
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"dataPath": "params_shard_24.bin",
|
| 466 |
+
"format": "raw-shard",
|
| 467 |
+
"nbytes": 50331648,
|
| 468 |
+
"records": [
|
| 469 |
+
{
|
| 470 |
+
"name": "model.layers.14.mlp.down_proj.weight",
|
| 471 |
+
"shape": [
|
| 472 |
+
3072,
|
| 473 |
+
8192
|
| 474 |
+
],
|
| 475 |
+
"dtype": "float16",
|
| 476 |
+
"format": "f32-to-bf16",
|
| 477 |
+
"nbytes": 50331648,
|
| 478 |
+
"byteOffset": 0
|
| 479 |
+
}
|
| 480 |
+
],
|
| 481 |
+
"md5sum": "dc2822cbc41bd102994668e4062c1490"
|
| 482 |
+
},
|
| 483 |
+
{
|
| 484 |
+
"dataPath": "params_shard_25.bin",
|
| 485 |
+
"format": "raw-shard",
|
| 486 |
+
"nbytes": 100663296,
|
| 487 |
+
"records": [
|
| 488 |
+
{
|
| 489 |
+
"name": "model.layers.14.mlp.gate_up_proj.weight",
|
| 490 |
+
"shape": [
|
| 491 |
+
16384,
|
| 492 |
+
3072
|
| 493 |
+
],
|
| 494 |
+
"dtype": "float16",
|
| 495 |
+
"format": "f32-to-bf16",
|
| 496 |
+
"nbytes": 100663296,
|
| 497 |
+
"byteOffset": 0
|
| 498 |
+
}
|
| 499 |
+
],
|
| 500 |
+
"md5sum": "119960688bc49d4410464d16f8d68b91"
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"dataPath": "params_shard_26.bin",
|
| 504 |
+
"format": "raw-shard",
|
| 505 |
+
"nbytes": 31457280,
|
| 506 |
+
"records": [
|
| 507 |
+
{
|
| 508 |
+
"name": "model.layers.14.self_attn.qkv_proj.weight",
|
| 509 |
+
"shape": [
|
| 510 |
+
5120,
|
| 511 |
+
3072
|
| 512 |
+
],
|
| 513 |
+
"dtype": "float16",
|
| 514 |
+
"format": "f32-to-bf16",
|
| 515 |
+
"nbytes": 31457280,
|
| 516 |
+
"byteOffset": 0
|
| 517 |
+
}
|
| 518 |
+
],
|
| 519 |
+
"md5sum": "ac6b7487ae103887d2e6cadb25a77bea"
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"dataPath": "params_shard_27.bin",
|
| 523 |
+
"format": "raw-shard",
|
| 524 |
+
"nbytes": 18874368,
|
| 525 |
+
"records": [
|
| 526 |
+
{
|
| 527 |
+
"name": "model.layers.14.self_attn.o_proj.weight",
|
| 528 |
+
"shape": [
|
| 529 |
+
3072,
|
| 530 |
+
3072
|
| 531 |
+
],
|
| 532 |
+
"dtype": "float16",
|
| 533 |
+
"format": "f32-to-bf16",
|
| 534 |
+
"nbytes": 18874368,
|
| 535 |
+
"byteOffset": 0
|
| 536 |
+
}
|
| 537 |
+
],
|
| 538 |
+
"md5sum": "cd5ff980cedce31715cb76142ca422f6"
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"dataPath": "params_shard_28.bin",
|
| 542 |
+
"format": "raw-shard",
|
| 543 |
+
"nbytes": 50331648,
|
| 544 |
+
"records": [
|
| 545 |
+
{
|
| 546 |
+
"name": "model.layers.15.mlp.down_proj.weight",
|
| 547 |
+
"shape": [
|
| 548 |
+
3072,
|
| 549 |
+
8192
|
| 550 |
+
],
|
| 551 |
+
"dtype": "float16",
|
| 552 |
+
"format": "f32-to-bf16",
|
| 553 |
+
"nbytes": 50331648,
|
| 554 |
+
"byteOffset": 0
|
| 555 |
+
}
|
| 556 |
+
],
|
| 557 |
+
"md5sum": "091631025e4b9ba90a61bf12172a16e6"
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"dataPath": "params_shard_29.bin",
|
| 561 |
+
"format": "raw-shard",
|
| 562 |
+
"nbytes": 100663296,
|
| 563 |
+
"records": [
|
| 564 |
+
{
|
| 565 |
+
"name": "model.layers.15.mlp.gate_up_proj.weight",
|
| 566 |
+
"shape": [
|
| 567 |
+
16384,
|
| 568 |
+
3072
|
| 569 |
+
],
|
| 570 |
+
"dtype": "float16",
|
| 571 |
+
"format": "f32-to-bf16",
|
| 572 |
+
"nbytes": 100663296,
|
| 573 |
+
"byteOffset": 0
|
| 574 |
+
}
|
| 575 |
+
],
|
| 576 |
+
"md5sum": "5f218d808438c1c5dc4f63f427657f51"
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"dataPath": "params_shard_30.bin",
|
| 580 |
+
"format": "raw-shard",
|
| 581 |
+
"nbytes": 31457280,
|
| 582 |
+
"records": [
|
| 583 |
+
{
|
| 584 |
+
"name": "model.layers.15.self_attn.qkv_proj.weight",
|
| 585 |
+
"shape": [
|
| 586 |
+
5120,
|
| 587 |
+
3072
|
| 588 |
+
],
|
| 589 |
+
"dtype": "float16",
|
| 590 |
+
"format": "f32-to-bf16",
|
| 591 |
+
"nbytes": 31457280,
|
| 592 |
+
"byteOffset": 0
|
| 593 |
+
}
|
| 594 |
+
],
|
| 595 |
+
"md5sum": "8ef192f6ac93f5bb799c0ef37d0e30e3"
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"dataPath": "params_shard_31.bin",
|
| 599 |
+
"format": "raw-shard",
|
| 600 |
+
"nbytes": 18874368,
|
| 601 |
+
"records": [
|
| 602 |
+
{
|
| 603 |
+
"name": "model.layers.15.self_attn.o_proj.weight",
|
| 604 |
+
"shape": [
|
| 605 |
+
3072,
|
| 606 |
+
3072
|
| 607 |
+
],
|
| 608 |
+
"dtype": "float16",
|
| 609 |
+
"format": "f32-to-bf16",
|
| 610 |
+
"nbytes": 18874368,
|
| 611 |
+
"byteOffset": 0
|
| 612 |
+
}
|
| 613 |
+
],
|
| 614 |
+
"md5sum": "d9168ee13fa059a41231ac8034456e6c"
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"dataPath": "params_shard_32.bin",
|
| 618 |
+
"format": "raw-shard",
|
| 619 |
+
"nbytes": 50331648,
|
| 620 |
+
"records": [
|
| 621 |
+
{
|
| 622 |
+
"name": "model.layers.16.mlp.down_proj.weight",
|
| 623 |
+
"shape": [
|
| 624 |
+
3072,
|
| 625 |
+
8192
|
| 626 |
+
],
|
| 627 |
+
"dtype": "float16",
|
| 628 |
+
"format": "f32-to-bf16",
|
| 629 |
+
"nbytes": 50331648,
|
| 630 |
+
"byteOffset": 0
|
| 631 |
+
}
|
| 632 |
+
],
|
| 633 |
+
"md5sum": "c3c5313db66e08011ce600c8808b355c"
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"dataPath": "params_shard_33.bin",
|
| 637 |
+
"format": "raw-shard",
|
| 638 |
+
"nbytes": 100663296,
|
| 639 |
+
"records": [
|
| 640 |
+
{
|
| 641 |
+
"name": "model.layers.16.mlp.gate_up_proj.weight",
|
| 642 |
+
"shape": [
|
| 643 |
+
16384,
|
| 644 |
+
3072
|
| 645 |
+
],
|
| 646 |
+
"dtype": "float16",
|
| 647 |
+
"format": "f32-to-bf16",
|
| 648 |
+
"nbytes": 100663296,
|
| 649 |
+
"byteOffset": 0
|
| 650 |
+
}
|
| 651 |
+
],
|
| 652 |
+
"md5sum": "a4f75c51e17fca235f346fc742144dbc"
|
| 653 |
+
},
|
| 654 |
+
{
|
| 655 |
+
"dataPath": "params_shard_34.bin",
|
| 656 |
+
"format": "raw-shard",
|
| 657 |
+
"nbytes": 31457280,
|
| 658 |
+
"records": [
|
| 659 |
+
{
|
| 660 |
+
"name": "model.layers.16.self_attn.qkv_proj.weight",
|
| 661 |
+
"shape": [
|
| 662 |
+
5120,
|
| 663 |
+
3072
|
| 664 |
+
],
|
| 665 |
+
"dtype": "float16",
|
| 666 |
+
"format": "f32-to-bf16",
|
| 667 |
+
"nbytes": 31457280,
|
| 668 |
+
"byteOffset": 0
|
| 669 |
+
}
|
| 670 |
+
],
|
| 671 |
+
"md5sum": "e3615902377476ed86467521108670ad"
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"dataPath": "params_shard_35.bin",
|
| 675 |
+
"format": "raw-shard",
|
| 676 |
+
"nbytes": 18874368,
|
| 677 |
+
"records": [
|
| 678 |
+
{
|
| 679 |
+
"name": "model.layers.16.self_attn.o_proj.weight",
|
| 680 |
+
"shape": [
|
| 681 |
+
3072,
|
| 682 |
+
3072
|
| 683 |
+
],
|
| 684 |
+
"dtype": "float16",
|
| 685 |
+
"format": "f32-to-bf16",
|
| 686 |
+
"nbytes": 18874368,
|
| 687 |
+
"byteOffset": 0
|
| 688 |
+
}
|
| 689 |
+
],
|
| 690 |
+
"md5sum": "bb81537fd44aa831afb3bfb5a1cddd17"
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"dataPath": "params_shard_36.bin",
|
| 694 |
+
"format": "raw-shard",
|
| 695 |
+
"nbytes": 50331648,
|
| 696 |
+
"records": [
|
| 697 |
+
{
|
| 698 |
+
"name": "model.layers.17.mlp.down_proj.weight",
|
| 699 |
+
"shape": [
|
| 700 |
+
3072,
|
| 701 |
+
8192
|
| 702 |
+
],
|
| 703 |
+
"dtype": "float16",
|
| 704 |
+
"format": "f32-to-bf16",
|
| 705 |
+
"nbytes": 50331648,
|
| 706 |
+
"byteOffset": 0
|
| 707 |
+
}
|
| 708 |
+
],
|
| 709 |
+
"md5sum": "9f0cc61b2b61536d436c80fc044cb49a"
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"dataPath": "params_shard_37.bin",
|
| 713 |
+
"format": "raw-shard",
|
| 714 |
+
"nbytes": 100663296,
|
| 715 |
+
"records": [
|
| 716 |
+
{
|
| 717 |
+
"name": "model.layers.17.mlp.gate_up_proj.weight",
|
| 718 |
+
"shape": [
|
| 719 |
+
16384,
|
| 720 |
+
3072
|
| 721 |
+
],
|
| 722 |
+
"dtype": "float16",
|
| 723 |
+
"format": "f32-to-bf16",
|
| 724 |
+
"nbytes": 100663296,
|
| 725 |
+
"byteOffset": 0
|
| 726 |
+
}
|
| 727 |
+
],
|
| 728 |
+
"md5sum": "96e577feaa61e6984b5868a7a70fdcdb"
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"dataPath": "params_shard_38.bin",
|
| 732 |
+
"format": "raw-shard",
|
| 733 |
+
"nbytes": 31457280,
|
| 734 |
+
"records": [
|
| 735 |
+
{
|
| 736 |
+
"name": "model.layers.17.self_attn.qkv_proj.weight",
|
| 737 |
+
"shape": [
|
| 738 |
+
5120,
|
| 739 |
+
3072
|
| 740 |
+
],
|
| 741 |
+
"dtype": "float16",
|
| 742 |
+
"format": "f32-to-bf16",
|
| 743 |
+
"nbytes": 31457280,
|
| 744 |
+
"byteOffset": 0
|
| 745 |
+
}
|
| 746 |
+
],
|
| 747 |
+
"md5sum": "fb31470bb7bb699ca815f6d487dc3c05"
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"dataPath": "params_shard_39.bin",
|
| 751 |
+
"format": "raw-shard",
|
| 752 |
+
"nbytes": 18874368,
|
| 753 |
+
"records": [
|
| 754 |
+
{
|
| 755 |
+
"name": "model.layers.17.self_attn.o_proj.weight",
|
| 756 |
+
"shape": [
|
| 757 |
+
3072,
|
| 758 |
+
3072
|
| 759 |
+
],
|
| 760 |
+
"dtype": "float16",
|
| 761 |
+
"format": "f32-to-bf16",
|
| 762 |
+
"nbytes": 18874368,
|
| 763 |
+
"byteOffset": 0
|
| 764 |
+
}
|
| 765 |
+
],
|
| 766 |
+
"md5sum": "b367cddcbe44c1864f11fc9ed83f96f3"
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"dataPath": "params_shard_40.bin",
|
| 770 |
+
"format": "raw-shard",
|
| 771 |
+
"nbytes": 50331648,
|
| 772 |
+
"records": [
|
| 773 |
+
{
|
| 774 |
+
"name": "model.layers.18.mlp.down_proj.weight",
|
| 775 |
+
"shape": [
|
| 776 |
+
3072,
|
| 777 |
+
8192
|
| 778 |
+
],
|
| 779 |
+
"dtype": "float16",
|
| 780 |
+
"format": "f32-to-bf16",
|
| 781 |
+
"nbytes": 50331648,
|
| 782 |
+
"byteOffset": 0
|
| 783 |
+
}
|
| 784 |
+
],
|
| 785 |
+
"md5sum": "e7cc7b53951b8844fa3b05b6a507f69a"
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"dataPath": "params_shard_41.bin",
|
| 789 |
+
"format": "raw-shard",
|
| 790 |
+
"nbytes": 100663296,
|
| 791 |
+
"records": [
|
| 792 |
+
{
|
| 793 |
+
"name": "model.layers.18.mlp.gate_up_proj.weight",
|
| 794 |
+
"shape": [
|
| 795 |
+
16384,
|
| 796 |
+
3072
|
| 797 |
+
],
|
| 798 |
+
"dtype": "float16",
|
| 799 |
+
"format": "f32-to-bf16",
|
| 800 |
+
"nbytes": 100663296,
|
| 801 |
+
"byteOffset": 0
|
| 802 |
+
}
|
| 803 |
+
],
|
| 804 |
+
"md5sum": "acb2012e225c8d875d2e6204506d4eac"
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"dataPath": "params_shard_42.bin",
|
| 808 |
+
"format": "raw-shard",
|
| 809 |
+
"nbytes": 31457280,
|
| 810 |
+
"records": [
|
| 811 |
+
{
|
| 812 |
+
"name": "model.layers.18.self_attn.qkv_proj.weight",
|
| 813 |
+
"shape": [
|
| 814 |
+
5120,
|
| 815 |
+
3072
|
| 816 |
+
],
|
| 817 |
+
"dtype": "float16",
|
| 818 |
+
"format": "f32-to-bf16",
|
| 819 |
+
"nbytes": 31457280,
|
| 820 |
+
"byteOffset": 0
|
| 821 |
+
}
|
| 822 |
+
],
|
| 823 |
+
"md5sum": "13283901e7a78ea5ca61433bc2641b25"
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"dataPath": "params_shard_43.bin",
|
| 827 |
+
"format": "raw-shard",
|
| 828 |
+
"nbytes": 18874368,
|
| 829 |
+
"records": [
|
| 830 |
+
{
|
| 831 |
+
"name": "model.layers.18.self_attn.o_proj.weight",
|
| 832 |
+
"shape": [
|
| 833 |
+
3072,
|
| 834 |
+
3072
|
| 835 |
+
],
|
| 836 |
+
"dtype": "float16",
|
| 837 |
+
"format": "f32-to-bf16",
|
| 838 |
+
"nbytes": 18874368,
|
| 839 |
+
"byteOffset": 0
|
| 840 |
+
}
|
| 841 |
+
],
|
| 842 |
+
"md5sum": "f3bc36db1a0a25ddb7103a428fd40e18"
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"dataPath": "params_shard_44.bin",
|
| 846 |
+
"format": "raw-shard",
|
| 847 |
+
"nbytes": 50331648,
|
| 848 |
+
"records": [
|
| 849 |
+
{
|
| 850 |
+
"name": "model.layers.19.mlp.down_proj.weight",
|
| 851 |
+
"shape": [
|
| 852 |
+
3072,
|
| 853 |
+
8192
|
| 854 |
+
],
|
| 855 |
+
"dtype": "float16",
|
| 856 |
+
"format": "f32-to-bf16",
|
| 857 |
+
"nbytes": 50331648,
|
| 858 |
+
"byteOffset": 0
|
| 859 |
+
}
|
| 860 |
+
],
|
| 861 |
+
"md5sum": "05a117aee8bd862976dced128cc6eb6b"
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"dataPath": "params_shard_45.bin",
|
| 865 |
+
"format": "raw-shard",
|
| 866 |
+
"nbytes": 100663296,
|
| 867 |
+
"records": [
|
| 868 |
+
{
|
| 869 |
+
"name": "model.layers.19.mlp.gate_up_proj.weight",
|
| 870 |
+
"shape": [
|
| 871 |
+
16384,
|
| 872 |
+
3072
|
| 873 |
+
],
|
| 874 |
+
"dtype": "float16",
|
| 875 |
+
"format": "f32-to-bf16",
|
| 876 |
+
"nbytes": 100663296,
|
| 877 |
+
"byteOffset": 0
|
| 878 |
+
}
|
| 879 |
+
],
|
| 880 |
+
"md5sum": "934e2cbd292621642feba80ac33786b8"
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"dataPath": "params_shard_46.bin",
|
| 884 |
+
"format": "raw-shard",
|
| 885 |
+
"nbytes": 31457280,
|
| 886 |
+
"records": [
|
| 887 |
+
{
|
| 888 |
+
"name": "model.layers.19.self_attn.qkv_proj.weight",
|
| 889 |
+
"shape": [
|
| 890 |
+
5120,
|
| 891 |
+
3072
|
| 892 |
+
],
|
| 893 |
+
"dtype": "float16",
|
| 894 |
+
"format": "f32-to-bf16",
|
| 895 |
+
"nbytes": 31457280,
|
| 896 |
+
"byteOffset": 0
|
| 897 |
+
}
|
| 898 |
+
],
|
| 899 |
+
"md5sum": "7eb4c5b3729ff22fe1162b8ef46d3ebf"
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"dataPath": "params_shard_47.bin",
|
| 903 |
+
"format": "raw-shard",
|
| 904 |
+
"nbytes": 18874368,
|
| 905 |
+
"records": [
|
| 906 |
+
{
|
| 907 |
+
"name": "model.layers.19.self_attn.o_proj.weight",
|
| 908 |
+
"shape": [
|
| 909 |
+
3072,
|
| 910 |
+
3072
|
| 911 |
+
],
|
| 912 |
+
"dtype": "float16",
|
| 913 |
+
"format": "f32-to-bf16",
|
| 914 |
+
"nbytes": 18874368,
|
| 915 |
+
"byteOffset": 0
|
| 916 |
+
}
|
| 917 |
+
],
|
| 918 |
+
"md5sum": "d047246a68038f9d8d65ef4608c0640c"
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"dataPath": "params_shard_48.bin",
|
| 922 |
+
"format": "raw-shard",
|
| 923 |
+
"nbytes": 50331648,
|
| 924 |
+
"records": [
|
| 925 |
+
{
|
| 926 |
+
"name": "model.layers.2.mlp.down_proj.weight",
|
| 927 |
+
"shape": [
|
| 928 |
+
3072,
|
| 929 |
+
8192
|
| 930 |
+
],
|
| 931 |
+
"dtype": "float16",
|
| 932 |
+
"format": "f32-to-bf16",
|
| 933 |
+
"nbytes": 50331648,
|
| 934 |
+
"byteOffset": 0
|
| 935 |
+
}
|
| 936 |
+
],
|
| 937 |
+
"md5sum": "52a7d358264e947cf3eebf73788cde0b"
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"dataPath": "params_shard_49.bin",
|
| 941 |
+
"format": "raw-shard",
|
| 942 |
+
"nbytes": 100663296,
|
| 943 |
+
"records": [
|
| 944 |
+
{
|
| 945 |
+
"name": "model.layers.2.mlp.gate_up_proj.weight",
|
| 946 |
+
"shape": [
|
| 947 |
+
16384,
|
| 948 |
+
3072
|
| 949 |
+
],
|
| 950 |
+
"dtype": "float16",
|
| 951 |
+
"format": "f32-to-bf16",
|
| 952 |
+
"nbytes": 100663296,
|
| 953 |
+
"byteOffset": 0
|
| 954 |
+
}
|
| 955 |
+
],
|
| 956 |
+
"md5sum": "1628743f1a42b9d5ec3fbda78089ffb8"
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"dataPath": "params_shard_50.bin",
|
| 960 |
+
"format": "raw-shard",
|
| 961 |
+
"nbytes": 31457280,
|
| 962 |
+
"records": [
|
| 963 |
+
{
|
| 964 |
+
"name": "model.layers.2.self_attn.qkv_proj.weight",
|
| 965 |
+
"shape": [
|
| 966 |
+
5120,
|
| 967 |
+
3072
|
| 968 |
+
],
|
| 969 |
+
"dtype": "float16",
|
| 970 |
+
"format": "f32-to-bf16",
|
| 971 |
+
"nbytes": 31457280,
|
| 972 |
+
"byteOffset": 0
|
| 973 |
+
}
|
| 974 |
+
],
|
| 975 |
+
"md5sum": "b3cd61d4bbdfcfa7f29c195f6ba2c473"
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"dataPath": "params_shard_51.bin",
|
| 979 |
+
"format": "raw-shard",
|
| 980 |
+
"nbytes": 18874368,
|
| 981 |
+
"records": [
|
| 982 |
+
{
|
| 983 |
+
"name": "model.layers.2.self_attn.o_proj.weight",
|
| 984 |
+
"shape": [
|
| 985 |
+
3072,
|
| 986 |
+
3072
|
| 987 |
+
],
|
| 988 |
+
"dtype": "float16",
|
| 989 |
+
"format": "f32-to-bf16",
|
| 990 |
+
"nbytes": 18874368,
|
| 991 |
+
"byteOffset": 0
|
| 992 |
+
}
|
| 993 |
+
],
|
| 994 |
+
"md5sum": "bc97d679bff31438e1a8890070ba5ac5"
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"dataPath": "params_shard_52.bin",
|
| 998 |
+
"format": "raw-shard",
|
| 999 |
+
"nbytes": 100663296,
|
| 1000 |
+
"records": [
|
| 1001 |
+
{
|
| 1002 |
+
"name": "model.layers.20.mlp.gate_up_proj.weight",
|
| 1003 |
+
"shape": [
|
| 1004 |
+
16384,
|
| 1005 |
+
3072
|
| 1006 |
+
],
|
| 1007 |
+
"dtype": "float16",
|
| 1008 |
+
"format": "f32-to-bf16",
|
| 1009 |
+
"nbytes": 100663296,
|
| 1010 |
+
"byteOffset": 0
|
| 1011 |
+
}
|
| 1012 |
+
],
|
| 1013 |
+
"md5sum": "ee225468a68e72348f7776fa30b0c3b9"
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"dataPath": "params_shard_53.bin",
|
| 1017 |
+
"format": "raw-shard",
|
| 1018 |
+
"nbytes": 31457280,
|
| 1019 |
+
"records": [
|
| 1020 |
+
{
|
| 1021 |
+
"name": "model.layers.20.self_attn.qkv_proj.weight",
|
| 1022 |
+
"shape": [
|
| 1023 |
+
5120,
|
| 1024 |
+
3072
|
| 1025 |
+
],
|
| 1026 |
+
"dtype": "float16",
|
| 1027 |
+
"format": "f32-to-bf16",
|
| 1028 |
+
"nbytes": 31457280,
|
| 1029 |
+
"byteOffset": 0
|
| 1030 |
+
}
|
| 1031 |
+
],
|
| 1032 |
+
"md5sum": "1d547401ec8fa9e8e2dd3bbcb8bb6ca1"
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"dataPath": "params_shard_54.bin",
|
| 1036 |
+
"format": "raw-shard",
|
| 1037 |
+
"nbytes": 18874368,
|
| 1038 |
+
"records": [
|
| 1039 |
+
{
|
| 1040 |
+
"name": "model.layers.20.self_attn.o_proj.weight",
|
| 1041 |
+
"shape": [
|
| 1042 |
+
3072,
|
| 1043 |
+
3072
|
| 1044 |
+
],
|
| 1045 |
+
"dtype": "float16",
|
| 1046 |
+
"format": "f32-to-bf16",
|
| 1047 |
+
"nbytes": 18874368,
|
| 1048 |
+
"byteOffset": 0
|
| 1049 |
+
}
|
| 1050 |
+
],
|
| 1051 |
+
"md5sum": "ad3cea57ffeab5dee2de4c3865b21f1e"
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"dataPath": "params_shard_55.bin",
|
| 1055 |
+
"format": "raw-shard",
|
| 1056 |
+
"nbytes": 50331648,
|
| 1057 |
+
"records": [
|
| 1058 |
+
{
|
| 1059 |
+
"name": "model.layers.3.mlp.down_proj.weight",
|
| 1060 |
+
"shape": [
|
| 1061 |
+
3072,
|
| 1062 |
+
8192
|
| 1063 |
+
],
|
| 1064 |
+
"dtype": "float16",
|
| 1065 |
+
"format": "f32-to-bf16",
|
| 1066 |
+
"nbytes": 50331648,
|
| 1067 |
+
"byteOffset": 0
|
| 1068 |
+
}
|
| 1069 |
+
],
|
| 1070 |
+
"md5sum": "6b44b2c4c18b97b68361b3a22ec98396"
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"dataPath": "params_shard_56.bin",
|
| 1074 |
+
"format": "raw-shard",
|
| 1075 |
+
"nbytes": 100663296,
|
| 1076 |
+
"records": [
|
| 1077 |
+
{
|
| 1078 |
+
"name": "model.layers.3.mlp.gate_up_proj.weight",
|
| 1079 |
+
"shape": [
|
| 1080 |
+
16384,
|
| 1081 |
+
3072
|
| 1082 |
+
],
|
| 1083 |
+
"dtype": "float16",
|
| 1084 |
+
"format": "f32-to-bf16",
|
| 1085 |
+
"nbytes": 100663296,
|
| 1086 |
+
"byteOffset": 0
|
| 1087 |
+
}
|
| 1088 |
+
],
|
| 1089 |
+
"md5sum": "5c78352a13928e4adb7aa99937ecdcc6"
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"dataPath": "params_shard_57.bin",
|
| 1093 |
+
"format": "raw-shard",
|
| 1094 |
+
"nbytes": 31457280,
|
| 1095 |
+
"records": [
|
| 1096 |
+
{
|
| 1097 |
+
"name": "model.layers.3.self_attn.qkv_proj.weight",
|
| 1098 |
+
"shape": [
|
| 1099 |
+
5120,
|
| 1100 |
+
3072
|
| 1101 |
+
],
|
| 1102 |
+
"dtype": "float16",
|
| 1103 |
+
"format": "f32-to-bf16",
|
| 1104 |
+
"nbytes": 31457280,
|
| 1105 |
+
"byteOffset": 0
|
| 1106 |
+
}
|
| 1107 |
+
],
|
| 1108 |
+
"md5sum": "d40b3feabdaaaacec05f41e53b6ad5f7"
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"dataPath": "params_shard_58.bin",
|
| 1112 |
+
"format": "raw-shard",
|
| 1113 |
+
"nbytes": 18874368,
|
| 1114 |
+
"records": [
|
| 1115 |
+
{
|
| 1116 |
+
"name": "model.layers.3.self_attn.o_proj.weight",
|
| 1117 |
+
"shape": [
|
| 1118 |
+
3072,
|
| 1119 |
+
3072
|
| 1120 |
+
],
|
| 1121 |
+
"dtype": "float16",
|
| 1122 |
+
"format": "f32-to-bf16",
|
| 1123 |
+
"nbytes": 18874368,
|
| 1124 |
+
"byteOffset": 0
|
| 1125 |
+
}
|
| 1126 |
+
],
|
| 1127 |
+
"md5sum": "8e1d27f24ffa0d7f834395567a0626cf"
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"dataPath": "params_shard_59.bin",
|
| 1131 |
+
"format": "raw-shard",
|
| 1132 |
+
"nbytes": 50331648,
|
| 1133 |
+
"records": [
|
| 1134 |
+
{
|
| 1135 |
+
"name": "model.layers.4.mlp.down_proj.weight",
|
| 1136 |
+
"shape": [
|
| 1137 |
+
3072,
|
| 1138 |
+
8192
|
| 1139 |
+
],
|
| 1140 |
+
"dtype": "float16",
|
| 1141 |
+
"format": "f32-to-bf16",
|
| 1142 |
+
"nbytes": 50331648,
|
| 1143 |
+
"byteOffset": 0
|
| 1144 |
+
}
|
| 1145 |
+
],
|
| 1146 |
+
"md5sum": "f5afb2097169c2c4a37777d8e78552c8"
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"dataPath": "params_shard_60.bin",
|
| 1150 |
+
"format": "raw-shard",
|
| 1151 |
+
"nbytes": 100663296,
|
| 1152 |
+
"records": [
|
| 1153 |
+
{
|
| 1154 |
+
"name": "model.layers.4.mlp.gate_up_proj.weight",
|
| 1155 |
+
"shape": [
|
| 1156 |
+
16384,
|
| 1157 |
+
3072
|
| 1158 |
+
],
|
| 1159 |
+
"dtype": "float16",
|
| 1160 |
+
"format": "f32-to-bf16",
|
| 1161 |
+
"nbytes": 100663296,
|
| 1162 |
+
"byteOffset": 0
|
| 1163 |
+
}
|
| 1164 |
+
],
|
| 1165 |
+
"md5sum": "4f83702709846ffc41b1d69452986319"
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"dataPath": "params_shard_61.bin",
|
| 1169 |
+
"format": "raw-shard",
|
| 1170 |
+
"nbytes": 31457280,
|
| 1171 |
+
"records": [
|
| 1172 |
+
{
|
| 1173 |
+
"name": "model.layers.4.self_attn.qkv_proj.weight",
|
| 1174 |
+
"shape": [
|
| 1175 |
+
5120,
|
| 1176 |
+
3072
|
| 1177 |
+
],
|
| 1178 |
+
"dtype": "float16",
|
| 1179 |
+
"format": "f32-to-bf16",
|
| 1180 |
+
"nbytes": 31457280,
|
| 1181 |
+
"byteOffset": 0
|
| 1182 |
+
}
|
| 1183 |
+
],
|
| 1184 |
+
"md5sum": "05e7664938e1a08cadecf5ba44d848c0"
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"dataPath": "params_shard_62.bin",
|
| 1188 |
+
"format": "raw-shard",
|
| 1189 |
+
"nbytes": 18874368,
|
| 1190 |
+
"records": [
|
| 1191 |
+
{
|
| 1192 |
+
"name": "model.layers.4.self_attn.o_proj.weight",
|
| 1193 |
+
"shape": [
|
| 1194 |
+
3072,
|
| 1195 |
+
3072
|
| 1196 |
+
],
|
| 1197 |
+
"dtype": "float16",
|
| 1198 |
+
"format": "f32-to-bf16",
|
| 1199 |
+
"nbytes": 18874368,
|
| 1200 |
+
"byteOffset": 0
|
| 1201 |
+
}
|
| 1202 |
+
],
|
| 1203 |
+
"md5sum": "5be6306bbe6b73e506de669f47c8b76c"
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"dataPath": "params_shard_63.bin",
|
| 1207 |
+
"format": "raw-shard",
|
| 1208 |
+
"nbytes": 50331648,
|
| 1209 |
+
"records": [
|
| 1210 |
+
{
|
| 1211 |
+
"name": "model.layers.5.mlp.down_proj.weight",
|
| 1212 |
+
"shape": [
|
| 1213 |
+
3072,
|
| 1214 |
+
8192
|
| 1215 |
+
],
|
| 1216 |
+
"dtype": "float16",
|
| 1217 |
+
"format": "f32-to-bf16",
|
| 1218 |
+
"nbytes": 50331648,
|
| 1219 |
+
"byteOffset": 0
|
| 1220 |
+
}
|
| 1221 |
+
],
|
| 1222 |
+
"md5sum": "a7be6ffe7e28f4b42cca0a640998d6b2"
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"dataPath": "params_shard_64.bin",
|
| 1226 |
+
"format": "raw-shard",
|
| 1227 |
+
"nbytes": 100663296,
|
| 1228 |
+
"records": [
|
| 1229 |
+
{
|
| 1230 |
+
"name": "model.layers.5.mlp.gate_up_proj.weight",
|
| 1231 |
+
"shape": [
|
| 1232 |
+
16384,
|
| 1233 |
+
3072
|
| 1234 |
+
],
|
| 1235 |
+
"dtype": "float16",
|
| 1236 |
+
"format": "f32-to-bf16",
|
| 1237 |
+
"nbytes": 100663296,
|
| 1238 |
+
"byteOffset": 0
|
| 1239 |
+
}
|
| 1240 |
+
],
|
| 1241 |
+
"md5sum": "3dddd3e32fc2a304c519753d9184e529"
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"dataPath": "params_shard_65.bin",
|
| 1245 |
+
"format": "raw-shard",
|
| 1246 |
+
"nbytes": 31457280,
|
| 1247 |
+
"records": [
|
| 1248 |
+
{
|
| 1249 |
+
"name": "model.layers.5.self_attn.qkv_proj.weight",
|
| 1250 |
+
"shape": [
|
| 1251 |
+
5120,
|
| 1252 |
+
3072
|
| 1253 |
+
],
|
| 1254 |
+
"dtype": "float16",
|
| 1255 |
+
"format": "f32-to-bf16",
|
| 1256 |
+
"nbytes": 31457280,
|
| 1257 |
+
"byteOffset": 0
|
| 1258 |
+
}
|
| 1259 |
+
],
|
| 1260 |
+
"md5sum": "461b504aaf3741f5ed568e8928094a6e"
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"dataPath": "params_shard_66.bin",
|
| 1264 |
+
"format": "raw-shard",
|
| 1265 |
+
"nbytes": 18874368,
|
| 1266 |
+
"records": [
|
| 1267 |
+
{
|
| 1268 |
+
"name": "model.layers.5.self_attn.o_proj.weight",
|
| 1269 |
+
"shape": [
|
| 1270 |
+
3072,
|
| 1271 |
+
3072
|
| 1272 |
+
],
|
| 1273 |
+
"dtype": "float16",
|
| 1274 |
+
"format": "f32-to-bf16",
|
| 1275 |
+
"nbytes": 18874368,
|
| 1276 |
+
"byteOffset": 0
|
| 1277 |
+
}
|
| 1278 |
+
],
|
| 1279 |
+
"md5sum": "26f9ffe723de0ede7825ff6b544c075a"
|
| 1280 |
+
},
|
| 1281 |
+
{
|
| 1282 |
+
"dataPath": "params_shard_67.bin",
|
| 1283 |
+
"format": "raw-shard",
|
| 1284 |
+
"nbytes": 50331648,
|
| 1285 |
+
"records": [
|
| 1286 |
+
{
|
| 1287 |
+
"name": "model.layers.6.mlp.down_proj.weight",
|
| 1288 |
+
"shape": [
|
| 1289 |
+
3072,
|
| 1290 |
+
8192
|
| 1291 |
+
],
|
| 1292 |
+
"dtype": "float16",
|
| 1293 |
+
"format": "f32-to-bf16",
|
| 1294 |
+
"nbytes": 50331648,
|
| 1295 |
+
"byteOffset": 0
|
| 1296 |
+
}
|
| 1297 |
+
],
|
| 1298 |
+
"md5sum": "4982aa8b1b1432d7dd8af059da328627"
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"dataPath": "params_shard_68.bin",
|
| 1302 |
+
"format": "raw-shard",
|
| 1303 |
+
"nbytes": 100663296,
|
| 1304 |
+
"records": [
|
| 1305 |
+
{
|
| 1306 |
+
"name": "model.layers.6.mlp.gate_up_proj.weight",
|
| 1307 |
+
"shape": [
|
| 1308 |
+
16384,
|
| 1309 |
+
3072
|
| 1310 |
+
],
|
| 1311 |
+
"dtype": "float16",
|
| 1312 |
+
"format": "f32-to-bf16",
|
| 1313 |
+
"nbytes": 100663296,
|
| 1314 |
+
"byteOffset": 0
|
| 1315 |
+
}
|
| 1316 |
+
],
|
| 1317 |
+
"md5sum": "7c90e16544155e069b6cdb5208c209c7"
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"dataPath": "params_shard_69.bin",
|
| 1321 |
+
"format": "raw-shard",
|
| 1322 |
+
"nbytes": 31457280,
|
| 1323 |
+
"records": [
|
| 1324 |
+
{
|
| 1325 |
+
"name": "model.layers.6.self_attn.qkv_proj.weight",
|
| 1326 |
+
"shape": [
|
| 1327 |
+
5120,
|
| 1328 |
+
3072
|
| 1329 |
+
],
|
| 1330 |
+
"dtype": "float16",
|
| 1331 |
+
"format": "f32-to-bf16",
|
| 1332 |
+
"nbytes": 31457280,
|
| 1333 |
+
"byteOffset": 0
|
| 1334 |
+
}
|
| 1335 |
+
],
|
| 1336 |
+
"md5sum": "d89651c2e38049b1a552c13627370e23"
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"dataPath": "params_shard_70.bin",
|
| 1340 |
+
"format": "raw-shard",
|
| 1341 |
+
"nbytes": 18874368,
|
| 1342 |
+
"records": [
|
| 1343 |
+
{
|
| 1344 |
+
"name": "model.layers.6.self_attn.o_proj.weight",
|
| 1345 |
+
"shape": [
|
| 1346 |
+
3072,
|
| 1347 |
+
3072
|
| 1348 |
+
],
|
| 1349 |
+
"dtype": "float16",
|
| 1350 |
+
"format": "f32-to-bf16",
|
| 1351 |
+
"nbytes": 18874368,
|
| 1352 |
+
"byteOffset": 0
|
| 1353 |
+
}
|
| 1354 |
+
],
|
| 1355 |
+
"md5sum": "ec306966c7e89f7895778ce8b14f4707"
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"dataPath": "params_shard_71.bin",
|
| 1359 |
+
"format": "raw-shard",
|
| 1360 |
+
"nbytes": 50331648,
|
| 1361 |
+
"records": [
|
| 1362 |
+
{
|
| 1363 |
+
"name": "model.layers.7.mlp.down_proj.weight",
|
| 1364 |
+
"shape": [
|
| 1365 |
+
3072,
|
| 1366 |
+
8192
|
| 1367 |
+
],
|
| 1368 |
+
"dtype": "float16",
|
| 1369 |
+
"format": "f32-to-bf16",
|
| 1370 |
+
"nbytes": 50331648,
|
| 1371 |
+
"byteOffset": 0
|
| 1372 |
+
}
|
| 1373 |
+
],
|
| 1374 |
+
"md5sum": "299c3c2258981caf33ed1d31cf8c78e2"
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"dataPath": "params_shard_72.bin",
|
| 1378 |
+
"format": "raw-shard",
|
| 1379 |
+
"nbytes": 100663296,
|
| 1380 |
+
"records": [
|
| 1381 |
+
{
|
| 1382 |
+
"name": "model.layers.7.mlp.gate_up_proj.weight",
|
| 1383 |
+
"shape": [
|
| 1384 |
+
16384,
|
| 1385 |
+
3072
|
| 1386 |
+
],
|
| 1387 |
+
"dtype": "float16",
|
| 1388 |
+
"format": "f32-to-bf16",
|
| 1389 |
+
"nbytes": 100663296,
|
| 1390 |
+
"byteOffset": 0
|
| 1391 |
+
}
|
| 1392 |
+
],
|
| 1393 |
+
"md5sum": "35d232bb60cbd45ca7a17a544c4dbf06"
|
| 1394 |
+
},
|
| 1395 |
+
{
|
| 1396 |
+
"dataPath": "params_shard_73.bin",
|
| 1397 |
+
"format": "raw-shard",
|
| 1398 |
+
"nbytes": 31457280,
|
| 1399 |
+
"records": [
|
| 1400 |
+
{
|
| 1401 |
+
"name": "model.layers.7.self_attn.qkv_proj.weight",
|
| 1402 |
+
"shape": [
|
| 1403 |
+
5120,
|
| 1404 |
+
3072
|
| 1405 |
+
],
|
| 1406 |
+
"dtype": "float16",
|
| 1407 |
+
"format": "f32-to-bf16",
|
| 1408 |
+
"nbytes": 31457280,
|
| 1409 |
+
"byteOffset": 0
|
| 1410 |
+
}
|
| 1411 |
+
],
|
| 1412 |
+
"md5sum": "4b3b50a160e7257b70ce5e2bbe6e3d6f"
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"dataPath": "params_shard_74.bin",
|
| 1416 |
+
"format": "raw-shard",
|
| 1417 |
+
"nbytes": 18874368,
|
| 1418 |
+
"records": [
|
| 1419 |
+
{
|
| 1420 |
+
"name": "model.layers.7.self_attn.o_proj.weight",
|
| 1421 |
+
"shape": [
|
| 1422 |
+
3072,
|
| 1423 |
+
3072
|
| 1424 |
+
],
|
| 1425 |
+
"dtype": "float16",
|
| 1426 |
+
"format": "f32-to-bf16",
|
| 1427 |
+
"nbytes": 18874368,
|
| 1428 |
+
"byteOffset": 0
|
| 1429 |
+
}
|
| 1430 |
+
],
|
| 1431 |
+
"md5sum": "20cdc2f04631e2a438cb6120031e62d9"
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"dataPath": "params_shard_75.bin",
|
| 1435 |
+
"format": "raw-shard",
|
| 1436 |
+
"nbytes": 50331648,
|
| 1437 |
+
"records": [
|
| 1438 |
+
{
|
| 1439 |
+
"name": "model.layers.8.mlp.down_proj.weight",
|
| 1440 |
+
"shape": [
|
| 1441 |
+
3072,
|
| 1442 |
+
8192
|
| 1443 |
+
],
|
| 1444 |
+
"dtype": "float16",
|
| 1445 |
+
"format": "f32-to-bf16",
|
| 1446 |
+
"nbytes": 50331648,
|
| 1447 |
+
"byteOffset": 0
|
| 1448 |
+
}
|
| 1449 |
+
],
|
| 1450 |
+
"md5sum": "5bf4e3aa941df2c0e7d3146a29510016"
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"dataPath": "params_shard_76.bin",
|
| 1454 |
+
"format": "raw-shard",
|
| 1455 |
+
"nbytes": 100663296,
|
| 1456 |
+
"records": [
|
| 1457 |
+
{
|
| 1458 |
+
"name": "model.layers.8.mlp.gate_up_proj.weight",
|
| 1459 |
+
"shape": [
|
| 1460 |
+
16384,
|
| 1461 |
+
3072
|
| 1462 |
+
],
|
| 1463 |
+
"dtype": "float16",
|
| 1464 |
+
"format": "f32-to-bf16",
|
| 1465 |
+
"nbytes": 100663296,
|
| 1466 |
+
"byteOffset": 0
|
| 1467 |
+
}
|
| 1468 |
+
],
|
| 1469 |
+
"md5sum": "293da23d33827869b58ac7f3ccd8d607"
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"dataPath": "params_shard_77.bin",
|
| 1473 |
+
"format": "raw-shard",
|
| 1474 |
+
"nbytes": 31457280,
|
| 1475 |
+
"records": [
|
| 1476 |
+
{
|
| 1477 |
+
"name": "model.layers.8.self_attn.qkv_proj.weight",
|
| 1478 |
+
"shape": [
|
| 1479 |
+
5120,
|
| 1480 |
+
3072
|
| 1481 |
+
],
|
| 1482 |
+
"dtype": "float16",
|
| 1483 |
+
"format": "f32-to-bf16",
|
| 1484 |
+
"nbytes": 31457280,
|
| 1485 |
+
"byteOffset": 0
|
| 1486 |
+
}
|
| 1487 |
+
],
|
| 1488 |
+
"md5sum": "b6a83e51dd363ec4fa410f01f6f406a3"
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"dataPath": "params_shard_78.bin",
|
| 1492 |
+
"format": "raw-shard",
|
| 1493 |
+
"nbytes": 18874368,
|
| 1494 |
+
"records": [
|
| 1495 |
+
{
|
| 1496 |
+
"name": "model.layers.8.self_attn.o_proj.weight",
|
| 1497 |
+
"shape": [
|
| 1498 |
+
3072,
|
| 1499 |
+
3072
|
| 1500 |
+
],
|
| 1501 |
+
"dtype": "float16",
|
| 1502 |
+
"format": "f32-to-bf16",
|
| 1503 |
+
"nbytes": 18874368,
|
| 1504 |
+
"byteOffset": 0
|
| 1505 |
+
}
|
| 1506 |
+
],
|
| 1507 |
+
"md5sum": "89043a4b43f7d9b52077c9bb7ebe1ced"
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"dataPath": "params_shard_79.bin",
|
| 1511 |
+
"format": "raw-shard",
|
| 1512 |
+
"nbytes": 50331648,
|
| 1513 |
+
"records": [
|
| 1514 |
+
{
|
| 1515 |
+
"name": "model.layers.9.mlp.down_proj.weight",
|
| 1516 |
+
"shape": [
|
| 1517 |
+
3072,
|
| 1518 |
+
8192
|
| 1519 |
+
],
|
| 1520 |
+
"dtype": "float16",
|
| 1521 |
+
"format": "f32-to-bf16",
|
| 1522 |
+
"nbytes": 50331648,
|
| 1523 |
+
"byteOffset": 0
|
| 1524 |
+
}
|
| 1525 |
+
],
|
| 1526 |
+
"md5sum": "fab351236a82a4ed4eaaef64c89fd857"
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"dataPath": "params_shard_80.bin",
|
| 1530 |
+
"format": "raw-shard",
|
| 1531 |
+
"nbytes": 100663296,
|
| 1532 |
+
"records": [
|
| 1533 |
+
{
|
| 1534 |
+
"name": "model.layers.9.mlp.gate_up_proj.weight",
|
| 1535 |
+
"shape": [
|
| 1536 |
+
16384,
|
| 1537 |
+
3072
|
| 1538 |
+
],
|
| 1539 |
+
"dtype": "float16",
|
| 1540 |
+
"format": "f32-to-bf16",
|
| 1541 |
+
"nbytes": 100663296,
|
| 1542 |
+
"byteOffset": 0
|
| 1543 |
+
}
|
| 1544 |
+
],
|
| 1545 |
+
"md5sum": "d997be401f1af829feb81876433f2e85"
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"dataPath": "params_shard_81.bin",
|
| 1549 |
+
"format": "raw-shard",
|
| 1550 |
+
"nbytes": 31457280,
|
| 1551 |
+
"records": [
|
| 1552 |
+
{
|
| 1553 |
+
"name": "model.layers.9.self_attn.qkv_proj.weight",
|
| 1554 |
+
"shape": [
|
| 1555 |
+
5120,
|
| 1556 |
+
3072
|
| 1557 |
+
],
|
| 1558 |
+
"dtype": "float16",
|
| 1559 |
+
"format": "f32-to-bf16",
|
| 1560 |
+
"nbytes": 31457280,
|
| 1561 |
+
"byteOffset": 0
|
| 1562 |
+
}
|
| 1563 |
+
],
|
| 1564 |
+
"md5sum": "6f56b9db892f385964dde0e37dee8e07"
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"dataPath": "params_shard_82.bin",
|
| 1568 |
+
"format": "raw-shard",
|
| 1569 |
+
"nbytes": 18874368,
|
| 1570 |
+
"records": [
|
| 1571 |
+
{
|
| 1572 |
+
"name": "model.layers.9.self_attn.o_proj.weight",
|
| 1573 |
+
"shape": [
|
| 1574 |
+
3072,
|
| 1575 |
+
3072
|
| 1576 |
+
],
|
| 1577 |
+
"dtype": "float16",
|
| 1578 |
+
"format": "f32-to-bf16",
|
| 1579 |
+
"nbytes": 18874368,
|
| 1580 |
+
"byteOffset": 0
|
| 1581 |
+
}
|
| 1582 |
+
],
|
| 1583 |
+
"md5sum": "6c853563e08a309c46b357b69d0374b4"
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"dataPath": "params_shard_83.bin",
|
| 1587 |
+
"format": "raw-shard",
|
| 1588 |
+
"nbytes": 50331648,
|
| 1589 |
+
"records": [
|
| 1590 |
+
{
|
| 1591 |
+
"name": "model.layers.20.mlp.down_proj.weight",
|
| 1592 |
+
"shape": [
|
| 1593 |
+
3072,
|
| 1594 |
+
8192
|
| 1595 |
+
],
|
| 1596 |
+
"dtype": "float16",
|
| 1597 |
+
"format": "f32-to-bf16",
|
| 1598 |
+
"nbytes": 50331648,
|
| 1599 |
+
"byteOffset": 0
|
| 1600 |
+
}
|
| 1601 |
+
],
|
| 1602 |
+
"md5sum": "edbfe726658fdea9d7ab1a2c72b9e127"
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"dataPath": "params_shard_84.bin",
|
| 1606 |
+
"format": "raw-shard",
|
| 1607 |
+
"nbytes": 50331648,
|
| 1608 |
+
"records": [
|
| 1609 |
+
{
|
| 1610 |
+
"name": "model.layers.21.mlp.down_proj.weight",
|
| 1611 |
+
"shape": [
|
| 1612 |
+
3072,
|
| 1613 |
+
8192
|
| 1614 |
+
],
|
| 1615 |
+
"dtype": "float16",
|
| 1616 |
+
"format": "f32-to-bf16",
|
| 1617 |
+
"nbytes": 50331648,
|
| 1618 |
+
"byteOffset": 0
|
| 1619 |
+
}
|
| 1620 |
+
],
|
| 1621 |
+
"md5sum": "8d409e8d7a30fbbaa4233b822ed0d300"
|
| 1622 |
+
},
|
| 1623 |
+
{
|
| 1624 |
+
"dataPath": "params_shard_85.bin",
|
| 1625 |
+
"format": "raw-shard",
|
| 1626 |
+
"nbytes": 100663296,
|
| 1627 |
+
"records": [
|
| 1628 |
+
{
|
| 1629 |
+
"name": "model.layers.21.mlp.gate_up_proj.weight",
|
| 1630 |
+
"shape": [
|
| 1631 |
+
16384,
|
| 1632 |
+
3072
|
| 1633 |
+
],
|
| 1634 |
+
"dtype": "float16",
|
| 1635 |
+
"format": "f32-to-bf16",
|
| 1636 |
+
"nbytes": 100663296,
|
| 1637 |
+
"byteOffset": 0
|
| 1638 |
+
}
|
| 1639 |
+
],
|
| 1640 |
+
"md5sum": "29ec9f881a83f51b7bdbb0b92a6226f2"
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"dataPath": "params_shard_86.bin",
|
| 1644 |
+
"format": "raw-shard",
|
| 1645 |
+
"nbytes": 31457280,
|
| 1646 |
+
"records": [
|
| 1647 |
+
{
|
| 1648 |
+
"name": "model.layers.21.self_attn.qkv_proj.weight",
|
| 1649 |
+
"shape": [
|
| 1650 |
+
5120,
|
| 1651 |
+
3072
|
| 1652 |
+
],
|
| 1653 |
+
"dtype": "float16",
|
| 1654 |
+
"format": "f32-to-bf16",
|
| 1655 |
+
"nbytes": 31457280,
|
| 1656 |
+
"byteOffset": 0
|
| 1657 |
+
}
|
| 1658 |
+
],
|
| 1659 |
+
"md5sum": "f80502b938f393358ef4350408bf094e"
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"dataPath": "params_shard_87.bin",
|
| 1663 |
+
"format": "raw-shard",
|
| 1664 |
+
"nbytes": 18874368,
|
| 1665 |
+
"records": [
|
| 1666 |
+
{
|
| 1667 |
+
"name": "model.layers.21.self_attn.o_proj.weight",
|
| 1668 |
+
"shape": [
|
| 1669 |
+
3072,
|
| 1670 |
+
3072
|
| 1671 |
+
],
|
| 1672 |
+
"dtype": "float16",
|
| 1673 |
+
"format": "f32-to-bf16",
|
| 1674 |
+
"nbytes": 18874368,
|
| 1675 |
+
"byteOffset": 0
|
| 1676 |
+
}
|
| 1677 |
+
],
|
| 1678 |
+
"md5sum": "f7e8db850638377b19c57093b00361ee"
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"dataPath": "params_shard_88.bin",
|
| 1682 |
+
"format": "raw-shard",
|
| 1683 |
+
"nbytes": 50331648,
|
| 1684 |
+
"records": [
|
| 1685 |
+
{
|
| 1686 |
+
"name": "model.layers.22.mlp.down_proj.weight",
|
| 1687 |
+
"shape": [
|
| 1688 |
+
3072,
|
| 1689 |
+
8192
|
| 1690 |
+
],
|
| 1691 |
+
"dtype": "float16",
|
| 1692 |
+
"format": "f32-to-bf16",
|
| 1693 |
+
"nbytes": 50331648,
|
| 1694 |
+
"byteOffset": 0
|
| 1695 |
+
}
|
| 1696 |
+
],
|
| 1697 |
+
"md5sum": "ce7ade97d9cb08c0411da69e046a3a7f"
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"dataPath": "params_shard_89.bin",
|
| 1701 |
+
"format": "raw-shard",
|
| 1702 |
+
"nbytes": 100663296,
|
| 1703 |
+
"records": [
|
| 1704 |
+
{
|
| 1705 |
+
"name": "model.layers.22.mlp.gate_up_proj.weight",
|
| 1706 |
+
"shape": [
|
| 1707 |
+
16384,
|
| 1708 |
+
3072
|
| 1709 |
+
],
|
| 1710 |
+
"dtype": "float16",
|
| 1711 |
+
"format": "f32-to-bf16",
|
| 1712 |
+
"nbytes": 100663296,
|
| 1713 |
+
"byteOffset": 0
|
| 1714 |
+
}
|
| 1715 |
+
],
|
| 1716 |
+
"md5sum": "59df560616188f1a16f02cd7e1497a71"
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"dataPath": "params_shard_90.bin",
|
| 1720 |
+
"format": "raw-shard",
|
| 1721 |
+
"nbytes": 31457280,
|
| 1722 |
+
"records": [
|
| 1723 |
+
{
|
| 1724 |
+
"name": "model.layers.22.self_attn.qkv_proj.weight",
|
| 1725 |
+
"shape": [
|
| 1726 |
+
5120,
|
| 1727 |
+
3072
|
| 1728 |
+
],
|
| 1729 |
+
"dtype": "float16",
|
| 1730 |
+
"format": "f32-to-bf16",
|
| 1731 |
+
"nbytes": 31457280,
|
| 1732 |
+
"byteOffset": 0
|
| 1733 |
+
}
|
| 1734 |
+
],
|
| 1735 |
+
"md5sum": "a4c7292ff72360c29e61c567c1443873"
|
| 1736 |
+
},
|
| 1737 |
+
{
|
| 1738 |
+
"dataPath": "params_shard_91.bin",
|
| 1739 |
+
"format": "raw-shard",
|
| 1740 |
+
"nbytes": 18874368,
|
| 1741 |
+
"records": [
|
| 1742 |
+
{
|
| 1743 |
+
"name": "model.layers.22.self_attn.o_proj.weight",
|
| 1744 |
+
"shape": [
|
| 1745 |
+
3072,
|
| 1746 |
+
3072
|
| 1747 |
+
],
|
| 1748 |
+
"dtype": "float16",
|
| 1749 |
+
"format": "f32-to-bf16",
|
| 1750 |
+
"nbytes": 18874368,
|
| 1751 |
+
"byteOffset": 0
|
| 1752 |
+
}
|
| 1753 |
+
],
|
| 1754 |
+
"md5sum": "c7029d83c52934cb8416529d37af541e"
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"dataPath": "params_shard_92.bin",
|
| 1758 |
+
"format": "raw-shard",
|
| 1759 |
+
"nbytes": 50331648,
|
| 1760 |
+
"records": [
|
| 1761 |
+
{
|
| 1762 |
+
"name": "model.layers.23.mlp.down_proj.weight",
|
| 1763 |
+
"shape": [
|
| 1764 |
+
3072,
|
| 1765 |
+
8192
|
| 1766 |
+
],
|
| 1767 |
+
"dtype": "float16",
|
| 1768 |
+
"format": "f32-to-bf16",
|
| 1769 |
+
"nbytes": 50331648,
|
| 1770 |
+
"byteOffset": 0
|
| 1771 |
+
}
|
| 1772 |
+
],
|
| 1773 |
+
"md5sum": "5a46e5db5af3fe6c4cacf434f13977d6"
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"dataPath": "params_shard_93.bin",
|
| 1777 |
+
"format": "raw-shard",
|
| 1778 |
+
"nbytes": 100663296,
|
| 1779 |
+
"records": [
|
| 1780 |
+
{
|
| 1781 |
+
"name": "model.layers.23.mlp.gate_up_proj.weight",
|
| 1782 |
+
"shape": [
|
| 1783 |
+
16384,
|
| 1784 |
+
3072
|
| 1785 |
+
],
|
| 1786 |
+
"dtype": "float16",
|
| 1787 |
+
"format": "f32-to-bf16",
|
| 1788 |
+
"nbytes": 100663296,
|
| 1789 |
+
"byteOffset": 0
|
| 1790 |
+
}
|
| 1791 |
+
],
|
| 1792 |
+
"md5sum": "824aa7d0fae4084ed9e4ba934127ef0b"
|
| 1793 |
+
},
|
| 1794 |
+
{
|
| 1795 |
+
"dataPath": "params_shard_94.bin",
|
| 1796 |
+
"format": "raw-shard",
|
| 1797 |
+
"nbytes": 31457280,
|
| 1798 |
+
"records": [
|
| 1799 |
+
{
|
| 1800 |
+
"name": "model.layers.23.self_attn.qkv_proj.weight",
|
| 1801 |
+
"shape": [
|
| 1802 |
+
5120,
|
| 1803 |
+
3072
|
| 1804 |
+
],
|
| 1805 |
+
"dtype": "float16",
|
| 1806 |
+
"format": "f32-to-bf16",
|
| 1807 |
+
"nbytes": 31457280,
|
| 1808 |
+
"byteOffset": 0
|
| 1809 |
+
}
|
| 1810 |
+
],
|
| 1811 |
+
"md5sum": "fcb306034f5325aaa26ac48c62fd16f9"
|
| 1812 |
+
},
|
| 1813 |
+
{
|
| 1814 |
+
"dataPath": "params_shard_95.bin",
|
| 1815 |
+
"format": "raw-shard",
|
| 1816 |
+
"nbytes": 18874368,
|
| 1817 |
+
"records": [
|
| 1818 |
+
{
|
| 1819 |
+
"name": "model.layers.23.self_attn.o_proj.weight",
|
| 1820 |
+
"shape": [
|
| 1821 |
+
3072,
|
| 1822 |
+
3072
|
| 1823 |
+
],
|
| 1824 |
+
"dtype": "float16",
|
| 1825 |
+
"format": "f32-to-bf16",
|
| 1826 |
+
"nbytes": 18874368,
|
| 1827 |
+
"byteOffset": 0
|
| 1828 |
+
}
|
| 1829 |
+
],
|
| 1830 |
+
"md5sum": "dc8109965ff7b0807bcd5ae3f48641c1"
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"dataPath": "params_shard_96.bin",
|
| 1834 |
+
"format": "raw-shard",
|
| 1835 |
+
"nbytes": 50331648,
|
| 1836 |
+
"records": [
|
| 1837 |
+
{
|
| 1838 |
+
"name": "model.layers.24.mlp.down_proj.weight",
|
| 1839 |
+
"shape": [
|
| 1840 |
+
3072,
|
| 1841 |
+
8192
|
| 1842 |
+
],
|
| 1843 |
+
"dtype": "float16",
|
| 1844 |
+
"format": "f32-to-bf16",
|
| 1845 |
+
"nbytes": 50331648,
|
| 1846 |
+
"byteOffset": 0
|
| 1847 |
+
}
|
| 1848 |
+
],
|
| 1849 |
+
"md5sum": "4406ea776d974f4423f84496deb69a49"
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"dataPath": "params_shard_97.bin",
|
| 1853 |
+
"format": "raw-shard",
|
| 1854 |
+
"nbytes": 100663296,
|
| 1855 |
+
"records": [
|
| 1856 |
+
{
|
| 1857 |
+
"name": "model.layers.24.mlp.gate_up_proj.weight",
|
| 1858 |
+
"shape": [
|
| 1859 |
+
16384,
|
| 1860 |
+
3072
|
| 1861 |
+
],
|
| 1862 |
+
"dtype": "float16",
|
| 1863 |
+
"format": "f32-to-bf16",
|
| 1864 |
+
"nbytes": 100663296,
|
| 1865 |
+
"byteOffset": 0
|
| 1866 |
+
}
|
| 1867 |
+
],
|
| 1868 |
+
"md5sum": "4adc35d6a727d2c2370ee6a26e369220"
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"dataPath": "params_shard_98.bin",
|
| 1872 |
+
"format": "raw-shard",
|
| 1873 |
+
"nbytes": 31457280,
|
| 1874 |
+
"records": [
|
| 1875 |
+
{
|
| 1876 |
+
"name": "model.layers.24.self_attn.qkv_proj.weight",
|
| 1877 |
+
"shape": [
|
| 1878 |
+
5120,
|
| 1879 |
+
3072
|
| 1880 |
+
],
|
| 1881 |
+
"dtype": "float16",
|
| 1882 |
+
"format": "f32-to-bf16",
|
| 1883 |
+
"nbytes": 31457280,
|
| 1884 |
+
"byteOffset": 0
|
| 1885 |
+
}
|
| 1886 |
+
],
|
| 1887 |
+
"md5sum": "8034e48ccafebfc3a7e0f7589f99222e"
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"dataPath": "params_shard_99.bin",
|
| 1891 |
+
"format": "raw-shard",
|
| 1892 |
+
"nbytes": 18874368,
|
| 1893 |
+
"records": [
|
| 1894 |
+
{
|
| 1895 |
+
"name": "model.layers.24.self_attn.o_proj.weight",
|
| 1896 |
+
"shape": [
|
| 1897 |
+
3072,
|
| 1898 |
+
3072
|
| 1899 |
+
],
|
| 1900 |
+
"dtype": "float16",
|
| 1901 |
+
"format": "f32-to-bf16",
|
| 1902 |
+
"nbytes": 18874368,
|
| 1903 |
+
"byteOffset": 0
|
| 1904 |
+
}
|
| 1905 |
+
],
|
| 1906 |
+
"md5sum": "0c3a596243236dd711ba13d5881234c8"
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"dataPath": "params_shard_100.bin",
|
| 1910 |
+
"format": "raw-shard",
|
| 1911 |
+
"nbytes": 50331648,
|
| 1912 |
+
"records": [
|
| 1913 |
+
{
|
| 1914 |
+
"name": "model.layers.25.mlp.down_proj.weight",
|
| 1915 |
+
"shape": [
|
| 1916 |
+
3072,
|
| 1917 |
+
8192
|
| 1918 |
+
],
|
| 1919 |
+
"dtype": "float16",
|
| 1920 |
+
"format": "f32-to-bf16",
|
| 1921 |
+
"nbytes": 50331648,
|
| 1922 |
+
"byteOffset": 0
|
| 1923 |
+
}
|
| 1924 |
+
],
|
| 1925 |
+
"md5sum": "36311bb7741c4ed38012a186f9e60eb0"
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"dataPath": "params_shard_101.bin",
|
| 1929 |
+
"format": "raw-shard",
|
| 1930 |
+
"nbytes": 100663296,
|
| 1931 |
+
"records": [
|
| 1932 |
+
{
|
| 1933 |
+
"name": "model.layers.25.mlp.gate_up_proj.weight",
|
| 1934 |
+
"shape": [
|
| 1935 |
+
16384,
|
| 1936 |
+
3072
|
| 1937 |
+
],
|
| 1938 |
+
"dtype": "float16",
|
| 1939 |
+
"format": "f32-to-bf16",
|
| 1940 |
+
"nbytes": 100663296,
|
| 1941 |
+
"byteOffset": 0
|
| 1942 |
+
}
|
| 1943 |
+
],
|
| 1944 |
+
"md5sum": "2c3ecd69de27229a3463f4055664ebd9"
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"dataPath": "params_shard_102.bin",
|
| 1948 |
+
"format": "raw-shard",
|
| 1949 |
+
"nbytes": 31457280,
|
| 1950 |
+
"records": [
|
| 1951 |
+
{
|
| 1952 |
+
"name": "model.layers.25.self_attn.qkv_proj.weight",
|
| 1953 |
+
"shape": [
|
| 1954 |
+
5120,
|
| 1955 |
+
3072
|
| 1956 |
+
],
|
| 1957 |
+
"dtype": "float16",
|
| 1958 |
+
"format": "f32-to-bf16",
|
| 1959 |
+
"nbytes": 31457280,
|
| 1960 |
+
"byteOffset": 0
|
| 1961 |
+
}
|
| 1962 |
+
],
|
| 1963 |
+
"md5sum": "e254ded35664683fb1b5cf12865cfce4"
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"dataPath": "params_shard_103.bin",
|
| 1967 |
+
"format": "raw-shard",
|
| 1968 |
+
"nbytes": 18874368,
|
| 1969 |
+
"records": [
|
| 1970 |
+
{
|
| 1971 |
+
"name": "model.layers.25.self_attn.o_proj.weight",
|
| 1972 |
+
"shape": [
|
| 1973 |
+
3072,
|
| 1974 |
+
3072
|
| 1975 |
+
],
|
| 1976 |
+
"dtype": "float16",
|
| 1977 |
+
"format": "f32-to-bf16",
|
| 1978 |
+
"nbytes": 18874368,
|
| 1979 |
+
"byteOffset": 0
|
| 1980 |
+
}
|
| 1981 |
+
],
|
| 1982 |
+
"md5sum": "0c679f6afc92093e60d6a57a9c3171fe"
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"dataPath": "params_shard_104.bin",
|
| 1986 |
+
"format": "raw-shard",
|
| 1987 |
+
"nbytes": 50331648,
|
| 1988 |
+
"records": [
|
| 1989 |
+
{
|
| 1990 |
+
"name": "model.layers.26.mlp.down_proj.weight",
|
| 1991 |
+
"shape": [
|
| 1992 |
+
3072,
|
| 1993 |
+
8192
|
| 1994 |
+
],
|
| 1995 |
+
"dtype": "float16",
|
| 1996 |
+
"format": "f32-to-bf16",
|
| 1997 |
+
"nbytes": 50331648,
|
| 1998 |
+
"byteOffset": 0
|
| 1999 |
+
}
|
| 2000 |
+
],
|
| 2001 |
+
"md5sum": "7770425eaa5ed60ca6aa257d46192768"
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"dataPath": "params_shard_105.bin",
|
| 2005 |
+
"format": "raw-shard",
|
| 2006 |
+
"nbytes": 100663296,
|
| 2007 |
+
"records": [
|
| 2008 |
+
{
|
| 2009 |
+
"name": "model.layers.26.mlp.gate_up_proj.weight",
|
| 2010 |
+
"shape": [
|
| 2011 |
+
16384,
|
| 2012 |
+
3072
|
| 2013 |
+
],
|
| 2014 |
+
"dtype": "float16",
|
| 2015 |
+
"format": "f32-to-bf16",
|
| 2016 |
+
"nbytes": 100663296,
|
| 2017 |
+
"byteOffset": 0
|
| 2018 |
+
}
|
| 2019 |
+
],
|
| 2020 |
+
"md5sum": "a3aae980f4e3945fa782a6e16f851048"
|
| 2021 |
+
},
|
| 2022 |
+
{
|
| 2023 |
+
"dataPath": "params_shard_106.bin",
|
| 2024 |
+
"format": "raw-shard",
|
| 2025 |
+
"nbytes": 31457280,
|
| 2026 |
+
"records": [
|
| 2027 |
+
{
|
| 2028 |
+
"name": "model.layers.26.self_attn.qkv_proj.weight",
|
| 2029 |
+
"shape": [
|
| 2030 |
+
5120,
|
| 2031 |
+
3072
|
| 2032 |
+
],
|
| 2033 |
+
"dtype": "float16",
|
| 2034 |
+
"format": "f32-to-bf16",
|
| 2035 |
+
"nbytes": 31457280,
|
| 2036 |
+
"byteOffset": 0
|
| 2037 |
+
}
|
| 2038 |
+
],
|
| 2039 |
+
"md5sum": "3273dfc4cead928697b5a50cb59829d4"
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"dataPath": "params_shard_107.bin",
|
| 2043 |
+
"format": "raw-shard",
|
| 2044 |
+
"nbytes": 18874368,
|
| 2045 |
+
"records": [
|
| 2046 |
+
{
|
| 2047 |
+
"name": "model.layers.26.self_attn.o_proj.weight",
|
| 2048 |
+
"shape": [
|
| 2049 |
+
3072,
|
| 2050 |
+
3072
|
| 2051 |
+
],
|
| 2052 |
+
"dtype": "float16",
|
| 2053 |
+
"format": "f32-to-bf16",
|
| 2054 |
+
"nbytes": 18874368,
|
| 2055 |
+
"byteOffset": 0
|
| 2056 |
+
}
|
| 2057 |
+
],
|
| 2058 |
+
"md5sum": "e03dcfad47c8923ee39be6b09fad64a2"
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"dataPath": "params_shard_108.bin",
|
| 2062 |
+
"format": "raw-shard",
|
| 2063 |
+
"nbytes": 50331648,
|
| 2064 |
+
"records": [
|
| 2065 |
+
{
|
| 2066 |
+
"name": "model.layers.27.mlp.down_proj.weight",
|
| 2067 |
+
"shape": [
|
| 2068 |
+
3072,
|
| 2069 |
+
8192
|
| 2070 |
+
],
|
| 2071 |
+
"dtype": "float16",
|
| 2072 |
+
"format": "f32-to-bf16",
|
| 2073 |
+
"nbytes": 50331648,
|
| 2074 |
+
"byteOffset": 0
|
| 2075 |
+
}
|
| 2076 |
+
],
|
| 2077 |
+
"md5sum": "a3c3a8a27fb7332cdd2b2af37deb6e1c"
|
| 2078 |
+
},
|
| 2079 |
+
{
|
| 2080 |
+
"dataPath": "params_shard_109.bin",
|
| 2081 |
+
"format": "raw-shard",
|
| 2082 |
+
"nbytes": 100663296,
|
| 2083 |
+
"records": [
|
| 2084 |
+
{
|
| 2085 |
+
"name": "model.layers.27.mlp.gate_up_proj.weight",
|
| 2086 |
+
"shape": [
|
| 2087 |
+
16384,
|
| 2088 |
+
3072
|
| 2089 |
+
],
|
| 2090 |
+
"dtype": "float16",
|
| 2091 |
+
"format": "f32-to-bf16",
|
| 2092 |
+
"nbytes": 100663296,
|
| 2093 |
+
"byteOffset": 0
|
| 2094 |
+
}
|
| 2095 |
+
],
|
| 2096 |
+
"md5sum": "bb9213dcffee9123b49a1172b2711600"
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"dataPath": "params_shard_110.bin",
|
| 2100 |
+
"format": "raw-shard",
|
| 2101 |
+
"nbytes": 31457280,
|
| 2102 |
+
"records": [
|
| 2103 |
+
{
|
| 2104 |
+
"name": "model.layers.27.self_attn.qkv_proj.weight",
|
| 2105 |
+
"shape": [
|
| 2106 |
+
5120,
|
| 2107 |
+
3072
|
| 2108 |
+
],
|
| 2109 |
+
"dtype": "float16",
|
| 2110 |
+
"format": "f32-to-bf16",
|
| 2111 |
+
"nbytes": 31457280,
|
| 2112 |
+
"byteOffset": 0
|
| 2113 |
+
}
|
| 2114 |
+
],
|
| 2115 |
+
"md5sum": "46c404a22b2c2db4bc1ace1b77e9aa57"
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"dataPath": "params_shard_111.bin",
|
| 2119 |
+
"format": "raw-shard",
|
| 2120 |
+
"nbytes": 18874368,
|
| 2121 |
+
"records": [
|
| 2122 |
+
{
|
| 2123 |
+
"name": "model.layers.27.self_attn.o_proj.weight",
|
| 2124 |
+
"shape": [
|
| 2125 |
+
3072,
|
| 2126 |
+
3072
|
| 2127 |
+
],
|
| 2128 |
+
"dtype": "float16",
|
| 2129 |
+
"format": "f32-to-bf16",
|
| 2130 |
+
"nbytes": 18874368,
|
| 2131 |
+
"byteOffset": 0
|
| 2132 |
+
}
|
| 2133 |
+
],
|
| 2134 |
+
"md5sum": "462fd0e9b2b626ee2144acfca5671c8c"
|
| 2135 |
+
},
|
| 2136 |
+
{
|
| 2137 |
+
"dataPath": "params_shard_112.bin",
|
| 2138 |
+
"format": "raw-shard",
|
| 2139 |
+
"nbytes": 31807488,
|
| 2140 |
+
"records": [
|
| 2141 |
+
{
|
| 2142 |
+
"name": "model.layers.0.input_layernorm.weight",
|
| 2143 |
+
"shape": [
|
| 2144 |
+
3072
|
| 2145 |
+
],
|
| 2146 |
+
"dtype": "float16",
|
| 2147 |
+
"format": "f32-to-bf16",
|
| 2148 |
+
"nbytes": 6144,
|
| 2149 |
+
"byteOffset": 0
|
| 2150 |
+
},
|
| 2151 |
+
{
|
| 2152 |
+
"name": "model.layers.0.post_attention_layernorm.weight",
|
| 2153 |
+
"shape": [
|
| 2154 |
+
3072
|
| 2155 |
+
],
|
| 2156 |
+
"dtype": "float16",
|
| 2157 |
+
"format": "f32-to-bf16",
|
| 2158 |
+
"nbytes": 6144,
|
| 2159 |
+
"byteOffset": 6144
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"name": "model.layers.0.self_attn.qkv_proj.weight",
|
| 2163 |
+
"shape": [
|
| 2164 |
+
5120,
|
| 2165 |
+
3072
|
| 2166 |
+
],
|
| 2167 |
+
"dtype": "float16",
|
| 2168 |
+
"format": "f32-to-bf16",
|
| 2169 |
+
"nbytes": 31457280,
|
| 2170 |
+
"byteOffset": 12288
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"name": "model.layers.1.input_layernorm.weight",
|
| 2174 |
+
"shape": [
|
| 2175 |
+
3072
|
| 2176 |
+
],
|
| 2177 |
+
"dtype": "float16",
|
| 2178 |
+
"format": "f32-to-bf16",
|
| 2179 |
+
"nbytes": 6144,
|
| 2180 |
+
"byteOffset": 31469568
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"name": "model.layers.1.post_attention_layernorm.weight",
|
| 2184 |
+
"shape": [
|
| 2185 |
+
3072
|
| 2186 |
+
],
|
| 2187 |
+
"dtype": "float16",
|
| 2188 |
+
"format": "f32-to-bf16",
|
| 2189 |
+
"nbytes": 6144,
|
| 2190 |
+
"byteOffset": 31475712
|
| 2191 |
+
},
|
| 2192 |
+
{
|
| 2193 |
+
"name": "model.layers.10.input_layernorm.weight",
|
| 2194 |
+
"shape": [
|
| 2195 |
+
3072
|
| 2196 |
+
],
|
| 2197 |
+
"dtype": "float16",
|
| 2198 |
+
"format": "f32-to-bf16",
|
| 2199 |
+
"nbytes": 6144,
|
| 2200 |
+
"byteOffset": 31481856
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"name": "model.layers.10.post_attention_layernorm.weight",
|
| 2204 |
+
"shape": [
|
| 2205 |
+
3072
|
| 2206 |
+
],
|
| 2207 |
+
"dtype": "float16",
|
| 2208 |
+
"format": "f32-to-bf16",
|
| 2209 |
+
"nbytes": 6144,
|
| 2210 |
+
"byteOffset": 31488000
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"name": "model.layers.11.input_layernorm.weight",
|
| 2214 |
+
"shape": [
|
| 2215 |
+
3072
|
| 2216 |
+
],
|
| 2217 |
+
"dtype": "float16",
|
| 2218 |
+
"format": "f32-to-bf16",
|
| 2219 |
+
"nbytes": 6144,
|
| 2220 |
+
"byteOffset": 31494144
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"name": "model.layers.11.post_attention_layernorm.weight",
|
| 2224 |
+
"shape": [
|
| 2225 |
+
3072
|
| 2226 |
+
],
|
| 2227 |
+
"dtype": "float16",
|
| 2228 |
+
"format": "f32-to-bf16",
|
| 2229 |
+
"nbytes": 6144,
|
| 2230 |
+
"byteOffset": 31500288
|
| 2231 |
+
},
|
| 2232 |
+
{
|
| 2233 |
+
"name": "model.layers.12.input_layernorm.weight",
|
| 2234 |
+
"shape": [
|
| 2235 |
+
3072
|
| 2236 |
+
],
|
| 2237 |
+
"dtype": "float16",
|
| 2238 |
+
"format": "f32-to-bf16",
|
| 2239 |
+
"nbytes": 6144,
|
| 2240 |
+
"byteOffset": 31506432
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"name": "model.layers.12.post_attention_layernorm.weight",
|
| 2244 |
+
"shape": [
|
| 2245 |
+
3072
|
| 2246 |
+
],
|
| 2247 |
+
"dtype": "float16",
|
| 2248 |
+
"format": "f32-to-bf16",
|
| 2249 |
+
"nbytes": 6144,
|
| 2250 |
+
"byteOffset": 31512576
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"name": "model.layers.13.input_layernorm.weight",
|
| 2254 |
+
"shape": [
|
| 2255 |
+
3072
|
| 2256 |
+
],
|
| 2257 |
+
"dtype": "float16",
|
| 2258 |
+
"format": "f32-to-bf16",
|
| 2259 |
+
"nbytes": 6144,
|
| 2260 |
+
"byteOffset": 31518720
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"name": "model.layers.13.post_attention_layernorm.weight",
|
| 2264 |
+
"shape": [
|
| 2265 |
+
3072
|
| 2266 |
+
],
|
| 2267 |
+
"dtype": "float16",
|
| 2268 |
+
"format": "f32-to-bf16",
|
| 2269 |
+
"nbytes": 6144,
|
| 2270 |
+
"byteOffset": 31524864
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"name": "model.layers.14.input_layernorm.weight",
|
| 2274 |
+
"shape": [
|
| 2275 |
+
3072
|
| 2276 |
+
],
|
| 2277 |
+
"dtype": "float16",
|
| 2278 |
+
"format": "f32-to-bf16",
|
| 2279 |
+
"nbytes": 6144,
|
| 2280 |
+
"byteOffset": 31531008
|
| 2281 |
+
},
|
| 2282 |
+
{
|
| 2283 |
+
"name": "model.layers.14.post_attention_layernorm.weight",
|
| 2284 |
+
"shape": [
|
| 2285 |
+
3072
|
| 2286 |
+
],
|
| 2287 |
+
"dtype": "float16",
|
| 2288 |
+
"format": "f32-to-bf16",
|
| 2289 |
+
"nbytes": 6144,
|
| 2290 |
+
"byteOffset": 31537152
|
| 2291 |
+
},
|
| 2292 |
+
{
|
| 2293 |
+
"name": "model.layers.15.input_layernorm.weight",
|
| 2294 |
+
"shape": [
|
| 2295 |
+
3072
|
| 2296 |
+
],
|
| 2297 |
+
"dtype": "float16",
|
| 2298 |
+
"format": "f32-to-bf16",
|
| 2299 |
+
"nbytes": 6144,
|
| 2300 |
+
"byteOffset": 31543296
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"name": "model.layers.15.post_attention_layernorm.weight",
|
| 2304 |
+
"shape": [
|
| 2305 |
+
3072
|
| 2306 |
+
],
|
| 2307 |
+
"dtype": "float16",
|
| 2308 |
+
"format": "f32-to-bf16",
|
| 2309 |
+
"nbytes": 6144,
|
| 2310 |
+
"byteOffset": 31549440
|
| 2311 |
+
},
|
| 2312 |
+
{
|
| 2313 |
+
"name": "model.layers.16.input_layernorm.weight",
|
| 2314 |
+
"shape": [
|
| 2315 |
+
3072
|
| 2316 |
+
],
|
| 2317 |
+
"dtype": "float16",
|
| 2318 |
+
"format": "f32-to-bf16",
|
| 2319 |
+
"nbytes": 6144,
|
| 2320 |
+
"byteOffset": 31555584
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"name": "model.layers.16.post_attention_layernorm.weight",
|
| 2324 |
+
"shape": [
|
| 2325 |
+
3072
|
| 2326 |
+
],
|
| 2327 |
+
"dtype": "float16",
|
| 2328 |
+
"format": "f32-to-bf16",
|
| 2329 |
+
"nbytes": 6144,
|
| 2330 |
+
"byteOffset": 31561728
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"name": "model.layers.17.input_layernorm.weight",
|
| 2334 |
+
"shape": [
|
| 2335 |
+
3072
|
| 2336 |
+
],
|
| 2337 |
+
"dtype": "float16",
|
| 2338 |
+
"format": "f32-to-bf16",
|
| 2339 |
+
"nbytes": 6144,
|
| 2340 |
+
"byteOffset": 31567872
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"name": "model.layers.17.post_attention_layernorm.weight",
|
| 2344 |
+
"shape": [
|
| 2345 |
+
3072
|
| 2346 |
+
],
|
| 2347 |
+
"dtype": "float16",
|
| 2348 |
+
"format": "f32-to-bf16",
|
| 2349 |
+
"nbytes": 6144,
|
| 2350 |
+
"byteOffset": 31574016
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"name": "model.layers.18.input_layernorm.weight",
|
| 2354 |
+
"shape": [
|
| 2355 |
+
3072
|
| 2356 |
+
],
|
| 2357 |
+
"dtype": "float16",
|
| 2358 |
+
"format": "f32-to-bf16",
|
| 2359 |
+
"nbytes": 6144,
|
| 2360 |
+
"byteOffset": 31580160
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"name": "model.layers.18.post_attention_layernorm.weight",
|
| 2364 |
+
"shape": [
|
| 2365 |
+
3072
|
| 2366 |
+
],
|
| 2367 |
+
"dtype": "float16",
|
| 2368 |
+
"format": "f32-to-bf16",
|
| 2369 |
+
"nbytes": 6144,
|
| 2370 |
+
"byteOffset": 31586304
|
| 2371 |
+
},
|
| 2372 |
+
{
|
| 2373 |
+
"name": "model.layers.19.input_layernorm.weight",
|
| 2374 |
+
"shape": [
|
| 2375 |
+
3072
|
| 2376 |
+
],
|
| 2377 |
+
"dtype": "float16",
|
| 2378 |
+
"format": "f32-to-bf16",
|
| 2379 |
+
"nbytes": 6144,
|
| 2380 |
+
"byteOffset": 31592448
|
| 2381 |
+
},
|
| 2382 |
+
{
|
| 2383 |
+
"name": "model.layers.19.post_attention_layernorm.weight",
|
| 2384 |
+
"shape": [
|
| 2385 |
+
3072
|
| 2386 |
+
],
|
| 2387 |
+
"dtype": "float16",
|
| 2388 |
+
"format": "f32-to-bf16",
|
| 2389 |
+
"nbytes": 6144,
|
| 2390 |
+
"byteOffset": 31598592
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"name": "model.layers.2.input_layernorm.weight",
|
| 2394 |
+
"shape": [
|
| 2395 |
+
3072
|
| 2396 |
+
],
|
| 2397 |
+
"dtype": "float16",
|
| 2398 |
+
"format": "f32-to-bf16",
|
| 2399 |
+
"nbytes": 6144,
|
| 2400 |
+
"byteOffset": 31604736
|
| 2401 |
+
},
|
| 2402 |
+
{
|
| 2403 |
+
"name": "model.layers.2.post_attention_layernorm.weight",
|
| 2404 |
+
"shape": [
|
| 2405 |
+
3072
|
| 2406 |
+
],
|
| 2407 |
+
"dtype": "float16",
|
| 2408 |
+
"format": "f32-to-bf16",
|
| 2409 |
+
"nbytes": 6144,
|
| 2410 |
+
"byteOffset": 31610880
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"name": "model.layers.3.input_layernorm.weight",
|
| 2414 |
+
"shape": [
|
| 2415 |
+
3072
|
| 2416 |
+
],
|
| 2417 |
+
"dtype": "float16",
|
| 2418 |
+
"format": "f32-to-bf16",
|
| 2419 |
+
"nbytes": 6144,
|
| 2420 |
+
"byteOffset": 31617024
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"name": "model.layers.3.post_attention_layernorm.weight",
|
| 2424 |
+
"shape": [
|
| 2425 |
+
3072
|
| 2426 |
+
],
|
| 2427 |
+
"dtype": "float16",
|
| 2428 |
+
"format": "f32-to-bf16",
|
| 2429 |
+
"nbytes": 6144,
|
| 2430 |
+
"byteOffset": 31623168
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"name": "model.layers.4.input_layernorm.weight",
|
| 2434 |
+
"shape": [
|
| 2435 |
+
3072
|
| 2436 |
+
],
|
| 2437 |
+
"dtype": "float16",
|
| 2438 |
+
"format": "f32-to-bf16",
|
| 2439 |
+
"nbytes": 6144,
|
| 2440 |
+
"byteOffset": 31629312
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"name": "model.layers.4.post_attention_layernorm.weight",
|
| 2444 |
+
"shape": [
|
| 2445 |
+
3072
|
| 2446 |
+
],
|
| 2447 |
+
"dtype": "float16",
|
| 2448 |
+
"format": "f32-to-bf16",
|
| 2449 |
+
"nbytes": 6144,
|
| 2450 |
+
"byteOffset": 31635456
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"name": "model.layers.5.input_layernorm.weight",
|
| 2454 |
+
"shape": [
|
| 2455 |
+
3072
|
| 2456 |
+
],
|
| 2457 |
+
"dtype": "float16",
|
| 2458 |
+
"format": "f32-to-bf16",
|
| 2459 |
+
"nbytes": 6144,
|
| 2460 |
+
"byteOffset": 31641600
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"name": "model.layers.5.post_attention_layernorm.weight",
|
| 2464 |
+
"shape": [
|
| 2465 |
+
3072
|
| 2466 |
+
],
|
| 2467 |
+
"dtype": "float16",
|
| 2468 |
+
"format": "f32-to-bf16",
|
| 2469 |
+
"nbytes": 6144,
|
| 2470 |
+
"byteOffset": 31647744
|
| 2471 |
+
},
|
| 2472 |
+
{
|
| 2473 |
+
"name": "model.layers.6.input_layernorm.weight",
|
| 2474 |
+
"shape": [
|
| 2475 |
+
3072
|
| 2476 |
+
],
|
| 2477 |
+
"dtype": "float16",
|
| 2478 |
+
"format": "f32-to-bf16",
|
| 2479 |
+
"nbytes": 6144,
|
| 2480 |
+
"byteOffset": 31653888
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"name": "model.layers.6.post_attention_layernorm.weight",
|
| 2484 |
+
"shape": [
|
| 2485 |
+
3072
|
| 2486 |
+
],
|
| 2487 |
+
"dtype": "float16",
|
| 2488 |
+
"format": "f32-to-bf16",
|
| 2489 |
+
"nbytes": 6144,
|
| 2490 |
+
"byteOffset": 31660032
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"name": "model.layers.7.input_layernorm.weight",
|
| 2494 |
+
"shape": [
|
| 2495 |
+
3072
|
| 2496 |
+
],
|
| 2497 |
+
"dtype": "float16",
|
| 2498 |
+
"format": "f32-to-bf16",
|
| 2499 |
+
"nbytes": 6144,
|
| 2500 |
+
"byteOffset": 31666176
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"name": "model.layers.7.post_attention_layernorm.weight",
|
| 2504 |
+
"shape": [
|
| 2505 |
+
3072
|
| 2506 |
+
],
|
| 2507 |
+
"dtype": "float16",
|
| 2508 |
+
"format": "f32-to-bf16",
|
| 2509 |
+
"nbytes": 6144,
|
| 2510 |
+
"byteOffset": 31672320
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"name": "model.layers.8.input_layernorm.weight",
|
| 2514 |
+
"shape": [
|
| 2515 |
+
3072
|
| 2516 |
+
],
|
| 2517 |
+
"dtype": "float16",
|
| 2518 |
+
"format": "f32-to-bf16",
|
| 2519 |
+
"nbytes": 6144,
|
| 2520 |
+
"byteOffset": 31678464
|
| 2521 |
+
},
|
| 2522 |
+
{
|
| 2523 |
+
"name": "model.layers.8.post_attention_layernorm.weight",
|
| 2524 |
+
"shape": [
|
| 2525 |
+
3072
|
| 2526 |
+
],
|
| 2527 |
+
"dtype": "float16",
|
| 2528 |
+
"format": "f32-to-bf16",
|
| 2529 |
+
"nbytes": 6144,
|
| 2530 |
+
"byteOffset": 31684608
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"name": "model.layers.9.input_layernorm.weight",
|
| 2534 |
+
"shape": [
|
| 2535 |
+
3072
|
| 2536 |
+
],
|
| 2537 |
+
"dtype": "float16",
|
| 2538 |
+
"format": "f32-to-bf16",
|
| 2539 |
+
"nbytes": 6144,
|
| 2540 |
+
"byteOffset": 31690752
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"name": "model.layers.9.post_attention_layernorm.weight",
|
| 2544 |
+
"shape": [
|
| 2545 |
+
3072
|
| 2546 |
+
],
|
| 2547 |
+
"dtype": "float16",
|
| 2548 |
+
"format": "f32-to-bf16",
|
| 2549 |
+
"nbytes": 6144,
|
| 2550 |
+
"byteOffset": 31696896
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"name": "model.layers.20.input_layernorm.weight",
|
| 2554 |
+
"shape": [
|
| 2555 |
+
3072
|
| 2556 |
+
],
|
| 2557 |
+
"dtype": "float16",
|
| 2558 |
+
"format": "f32-to-bf16",
|
| 2559 |
+
"nbytes": 6144,
|
| 2560 |
+
"byteOffset": 31703040
|
| 2561 |
+
},
|
| 2562 |
+
{
|
| 2563 |
+
"name": "model.layers.20.post_attention_layernorm.weight",
|
| 2564 |
+
"shape": [
|
| 2565 |
+
3072
|
| 2566 |
+
],
|
| 2567 |
+
"dtype": "float16",
|
| 2568 |
+
"format": "f32-to-bf16",
|
| 2569 |
+
"nbytes": 6144,
|
| 2570 |
+
"byteOffset": 31709184
|
| 2571 |
+
},
|
| 2572 |
+
{
|
| 2573 |
+
"name": "model.layers.21.input_layernorm.weight",
|
| 2574 |
+
"shape": [
|
| 2575 |
+
3072
|
| 2576 |
+
],
|
| 2577 |
+
"dtype": "float16",
|
| 2578 |
+
"format": "f32-to-bf16",
|
| 2579 |
+
"nbytes": 6144,
|
| 2580 |
+
"byteOffset": 31715328
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"name": "model.layers.21.post_attention_layernorm.weight",
|
| 2584 |
+
"shape": [
|
| 2585 |
+
3072
|
| 2586 |
+
],
|
| 2587 |
+
"dtype": "float16",
|
| 2588 |
+
"format": "f32-to-bf16",
|
| 2589 |
+
"nbytes": 6144,
|
| 2590 |
+
"byteOffset": 31721472
|
| 2591 |
+
},
|
| 2592 |
+
{
|
| 2593 |
+
"name": "model.layers.22.input_layernorm.weight",
|
| 2594 |
+
"shape": [
|
| 2595 |
+
3072
|
| 2596 |
+
],
|
| 2597 |
+
"dtype": "float16",
|
| 2598 |
+
"format": "f32-to-bf16",
|
| 2599 |
+
"nbytes": 6144,
|
| 2600 |
+
"byteOffset": 31727616
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"name": "model.layers.22.post_attention_layernorm.weight",
|
| 2604 |
+
"shape": [
|
| 2605 |
+
3072
|
| 2606 |
+
],
|
| 2607 |
+
"dtype": "float16",
|
| 2608 |
+
"format": "f32-to-bf16",
|
| 2609 |
+
"nbytes": 6144,
|
| 2610 |
+
"byteOffset": 31733760
|
| 2611 |
+
},
|
| 2612 |
+
{
|
| 2613 |
+
"name": "model.layers.23.input_layernorm.weight",
|
| 2614 |
+
"shape": [
|
| 2615 |
+
3072
|
| 2616 |
+
],
|
| 2617 |
+
"dtype": "float16",
|
| 2618 |
+
"format": "f32-to-bf16",
|
| 2619 |
+
"nbytes": 6144,
|
| 2620 |
+
"byteOffset": 31739904
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"name": "model.layers.23.post_attention_layernorm.weight",
|
| 2624 |
+
"shape": [
|
| 2625 |
+
3072
|
| 2626 |
+
],
|
| 2627 |
+
"dtype": "float16",
|
| 2628 |
+
"format": "f32-to-bf16",
|
| 2629 |
+
"nbytes": 6144,
|
| 2630 |
+
"byteOffset": 31746048
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"name": "model.layers.24.input_layernorm.weight",
|
| 2634 |
+
"shape": [
|
| 2635 |
+
3072
|
| 2636 |
+
],
|
| 2637 |
+
"dtype": "float16",
|
| 2638 |
+
"format": "f32-to-bf16",
|
| 2639 |
+
"nbytes": 6144,
|
| 2640 |
+
"byteOffset": 31752192
|
| 2641 |
+
},
|
| 2642 |
+
{
|
| 2643 |
+
"name": "model.layers.24.post_attention_layernorm.weight",
|
| 2644 |
+
"shape": [
|
| 2645 |
+
3072
|
| 2646 |
+
],
|
| 2647 |
+
"dtype": "float16",
|
| 2648 |
+
"format": "f32-to-bf16",
|
| 2649 |
+
"nbytes": 6144,
|
| 2650 |
+
"byteOffset": 31758336
|
| 2651 |
+
},
|
| 2652 |
+
{
|
| 2653 |
+
"name": "model.layers.25.input_layernorm.weight",
|
| 2654 |
+
"shape": [
|
| 2655 |
+
3072
|
| 2656 |
+
],
|
| 2657 |
+
"dtype": "float16",
|
| 2658 |
+
"format": "f32-to-bf16",
|
| 2659 |
+
"nbytes": 6144,
|
| 2660 |
+
"byteOffset": 31764480
|
| 2661 |
+
},
|
| 2662 |
+
{
|
| 2663 |
+
"name": "model.layers.25.post_attention_layernorm.weight",
|
| 2664 |
+
"shape": [
|
| 2665 |
+
3072
|
| 2666 |
+
],
|
| 2667 |
+
"dtype": "float16",
|
| 2668 |
+
"format": "f32-to-bf16",
|
| 2669 |
+
"nbytes": 6144,
|
| 2670 |
+
"byteOffset": 31770624
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"name": "model.layers.26.input_layernorm.weight",
|
| 2674 |
+
"shape": [
|
| 2675 |
+
3072
|
| 2676 |
+
],
|
| 2677 |
+
"dtype": "float16",
|
| 2678 |
+
"format": "f32-to-bf16",
|
| 2679 |
+
"nbytes": 6144,
|
| 2680 |
+
"byteOffset": 31776768
|
| 2681 |
+
},
|
| 2682 |
+
{
|
| 2683 |
+
"name": "model.layers.26.post_attention_layernorm.weight",
|
| 2684 |
+
"shape": [
|
| 2685 |
+
3072
|
| 2686 |
+
],
|
| 2687 |
+
"dtype": "float16",
|
| 2688 |
+
"format": "f32-to-bf16",
|
| 2689 |
+
"nbytes": 6144,
|
| 2690 |
+
"byteOffset": 31782912
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"name": "model.layers.27.input_layernorm.weight",
|
| 2694 |
+
"shape": [
|
| 2695 |
+
3072
|
| 2696 |
+
],
|
| 2697 |
+
"dtype": "float16",
|
| 2698 |
+
"format": "f32-to-bf16",
|
| 2699 |
+
"nbytes": 6144,
|
| 2700 |
+
"byteOffset": 31789056
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"name": "model.layers.27.post_attention_layernorm.weight",
|
| 2704 |
+
"shape": [
|
| 2705 |
+
3072
|
| 2706 |
+
],
|
| 2707 |
+
"dtype": "float16",
|
| 2708 |
+
"format": "f32-to-bf16",
|
| 2709 |
+
"nbytes": 6144,
|
| 2710 |
+
"byteOffset": 31795200
|
| 2711 |
+
},
|
| 2712 |
+
{
|
| 2713 |
+
"name": "model.norm.weight",
|
| 2714 |
+
"shape": [
|
| 2715 |
+
3072
|
| 2716 |
+
],
|
| 2717 |
+
"dtype": "float16",
|
| 2718 |
+
"format": "f32-to-bf16",
|
| 2719 |
+
"nbytes": 6144,
|
| 2720 |
+
"byteOffset": 31801344
|
| 2721 |
+
}
|
| 2722 |
+
],
|
| 2723 |
+
"md5sum": "8ee51d640f3008dcccc13eac7d6b36b0"
|
| 2724 |
+
}
|
| 2725 |
+
]
|
| 2726 |
+
}
|
params_shard_0.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3108f17714b17ea1bd0a63ba24cdbf9aa67bd41ccfed20e486bad840e7446ba0
|
| 3 |
+
size 788004864
|
params_shard_1.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5c36ad9d7e1bdd4cee895e483ee7d9e271438806bbaf289ddc4a2b9b133d5127
|
| 3 |
+
size 50331648
|
params_shard_10.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f4f7462910988825789b327b5307aa5d3c9d5f9cd7c70ce8fa76fc9ba71fca2b
|
| 3 |
+
size 31457280
|
params_shard_100.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e9b480aa5daadcf3ec0520e4b5eb22cafcafe936c84e47485c526460685df432
|
| 3 |
+
size 50331648
|
params_shard_101.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:892a36a7c8830b2db5568421d52acc5107c502491278d5fd441399407dc8469c
|
| 3 |
+
size 100663296
|
params_shard_102.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ce526264e4a212d210e924bc25cea7fa5bf66dd6e7016922c37d02571c351d03
|
| 3 |
+
size 31457280
|
params_shard_103.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eec4341535982635c996e05275168402c15d4e7dea0e4610cea84994033d4600
|
| 3 |
+
size 18874368
|
params_shard_104.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:938467714d7f6786693360e973fe993d943d1946bb54d11d89e9c9d91ac0de09
|
| 3 |
+
size 50331648
|
params_shard_105.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e5aff105275807be4841cae33bebcf340b761fe292398de533925ee7eb044cf8
|
| 3 |
+
size 100663296
|
params_shard_106.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c4c6bd6e01325acbd740a3410426c54acb42420b4a14cd489049404b59a4ac0b
|
| 3 |
+
size 31457280
|
params_shard_107.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6f9559a11b5353b07bb710366422d9e42e747a49b30b7022b14af91b142e4653
|
| 3 |
+
size 18874368
|
params_shard_108.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:765fb60e22faabfc71f3dfa96b7060099b29076c7dc9ff0f39fe71d6dffd08d1
|
| 3 |
+
size 50331648
|
params_shard_109.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4d539534485bd4339f048d310887df03be2532886ea9cb7f098ae3836a364044
|
| 3 |
+
size 100663296
|
params_shard_11.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:db5ffe508c2c3aa9d210a8dfb43e26964d8d7aee8cc4b36d67cb522cda3eb29d
|
| 3 |
+
size 18874368
|
params_shard_110.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:80bb6166a11fc869ce8255cbbdee5a2f062bd92d9eab8f94a41833b41cc66269
|
| 3 |
+
size 31457280
|
params_shard_111.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ca7be06567749564b4a0d8c97955ba025396550afb1f4043c90c61224501b1e3
|
| 3 |
+
size 18874368
|
params_shard_112.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1c0168381da5fd64bd50f8bd8df4e6d483b6b96ec17a35ecf777db87c6e104c9
|
| 3 |
+
size 31807488
|
params_shard_12.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e6d4992d320250ff4cd276e337d41d7df2d045083d0a3d23fbcc529ff2ec16e1
|
| 3 |
+
size 50331648
|
params_shard_13.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9100c2e413a03de1218e3bbb933258f6be5a4b4146d4b5dd5f33b19ef3eeea1e
|
| 3 |
+
size 100663296
|
params_shard_14.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:080a20973d57a6459235ca509c98fad1b68ef7f5016e6e3063d799c213716852
|
| 3 |
+
size 31457280
|
params_shard_15.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:950816f170f968ea36b18aa4e70e93cf643289f53ef29810ee19f36a531a44ec
|
| 3 |
+
size 18874368
|
params_shard_16.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:48de39077d4b87728b1811f8d42cb0ae8b1ef9d68aa253b1b1c3578ade4811c2
|
| 3 |
+
size 50331648
|
params_shard_17.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:29809c7d06202e414262215c1db12eff7a492b9b137314a3fe9b141d9d58df26
|
| 3 |
+
size 100663296
|
params_shard_18.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2e46deeaf7088a45085a8e52d9763b2e08104a1b1bab7cd6e3abada804f33b7b
|
| 3 |
+
size 31457280
|
params_shard_19.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:85db630f79d6e86b0146e721cdb7160ec0e96bb13409d1a712e7b4c8e996efba
|
| 3 |
+
size 18874368
|
params_shard_2.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d4352a12dda320490149c3a6d898977f31c2eca88a6a2198f1deb125b2a99b87
|
| 3 |
+
size 100663296
|
params_shard_20.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c7e8c9a189465694e61c364c6ace7d936f2889fd5cc80bfa52ae70b0c95ea154
|
| 3 |
+
size 50331648
|
params_shard_21.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9db529d59610fcad5fded9ef4e39bb4e9b2846c1218263804b6842c20c89b993
|
| 3 |
+
size 100663296
|
params_shard_22.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4af52ff2dc8edabf53a30c63f429c77f7429307569bcae9ddf36818cec9b3352
|
| 3 |
+
size 31457280
|
params_shard_23.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4e8285d9e1aae22fcb26ffce7f47b43031f9481c69787d74ef6af76f962d2fa1
|
| 3 |
+
size 18874368
|
params_shard_24.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a7826eb202125323e3c5bf97447f50015465b73da4fcd29429d4687b49f90c35
|
| 3 |
+
size 50331648
|
params_shard_25.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:712f0545ed7386470dcf0735a4fa69c78239d5e8a6b65cc4b203bdb97917d215
|
| 3 |
+
size 100663296
|
params_shard_26.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9b17dff790c49f3dec30eeda5767d517152180ce9401f53b13c5b83f97e26a40
|
| 3 |
+
size 31457280
|
params_shard_27.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b7e6050646ed7478bc1f5c7db9c8d65aaf61a4a5c25449b752cd7a94eafec340
|
| 3 |
+
size 18874368
|
params_shard_28.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:57de52773f43ff0b261f853e2745e41b006fed76763028334d95d746a2f94f83
|
| 3 |
+
size 50331648
|
params_shard_29.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:52dd5654c66984d09e1461f3ab27724ed22cafb118432703fb4d1a5874498e7e
|
| 3 |
+
size 100663296
|
params_shard_3.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0b4d6999513604cdb95e19f3d04d63befe0acfb2849f8d37c959d6ca470554b9
|
| 3 |
+
size 18874368
|
params_shard_30.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ca28dbd279ed3310cd93af0e3b3647997062a9015f3998f8bdb9fe6834ee6674
|
| 3 |
+
size 31457280
|
params_shard_31.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:556b1b16d4e5f1c5e5117bf555989ac0a1cffcdb2f1b1204b3cc3597d13920fa
|
| 3 |
+
size 18874368
|
params_shard_32.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cc8a086e959127faaf804017b9cb349f29804d870351d583d768d85d59abe766
|
| 3 |
+
size 50331648
|
params_shard_33.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bd112ee0cc9416c2256dc4e8ff23fc6d4fac5b470d58c29b87801e28d670506d
|
| 3 |
+
size 100663296
|
params_shard_34.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dc680708c110630934c593386d32870aeca2ca5d762bdccce13fceb5980de1b0
|
| 3 |
+
size 31457280
|
params_shard_35.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:012cefc6ef80029f128cbab7420f13df84804ebd7b857029e43f479a591d304a
|
| 3 |
+
size 18874368
|
params_shard_36.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:13ec6325053c7c588971fd16c58fe39a1feffc143f89c41bba0b6bc431388cfe
|
| 3 |
+
size 50331648
|
params_shard_37.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0f309dabf1a52633eda88c9e97dc7cb8553eb833ecac9a9a70a04f94e1f160c5
|
| 3 |
+
size 100663296
|
params_shard_38.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5613ae519725a52aa4e651348fffee1c61ba58422e1b7ea1f791619cf6c92952
|
| 3 |
+
size 31457280
|
params_shard_39.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7810af9518a9c1eeb9a45e2cbc4713aa0598497fbd06467cf5e4394834aabaf5
|
| 3 |
+
size 18874368
|
params_shard_4.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:96c48f7f16d264fa6c80790fc354a6b42ed20c51a0b8aa5b598c654d1c3c2def
|
| 3 |
+
size 50331648
|