Hugging Face's logo

Chat-Error
/
Blind-test01

Model card Files
xet
Community
nRuaif commited on
Commit
c4fc286
·
1 Parent(s): 846dde2

Upload folder using huggingface_hub

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .ipynb_checkpoints/Untitled-checkpoint.ipynb +6 -0
  2. Untitled.ipynb +50 -0
  3. adapter_config.json +26 -0
  4. added_tokens.json +3 -0
  5. checkpoint-10/README.md +34 -0
  6. checkpoint-10/adapter_config.json +26 -0
  7. checkpoint-10/adapter_model.bin +3 -0
  8. checkpoint-10/adapter_model/README.md +34 -0
  9. checkpoint-10/adapter_model/adapter_config.json +26 -0
  10. checkpoint-10/adapter_model/adapter_model.bin +3 -0
  11. checkpoint-10/optimizer.pt +3 -0
  12. checkpoint-10/rng_state.pth +3 -0
  13. checkpoint-10/scheduler.pt +3 -0
  14. checkpoint-10/trainer_state.json +87 -0
  15. checkpoint-10/training_args.bin +3 -0
  16. checkpoint-20/README.md +21 -0
  17. checkpoint-20/adapter_config.json +26 -0
  18. checkpoint-20/adapter_model.bin +3 -0
  19. checkpoint-20/adapter_model/README.md +21 -0
  20. checkpoint-20/adapter_model/adapter_config.json +26 -0
  21. checkpoint-20/adapter_model/adapter_model.bin +3 -0
  22. checkpoint-20/optimizer.pt +3 -0
  23. checkpoint-20/rng_state.pth +3 -0
  24. checkpoint-20/scheduler.pt +3 -0
  25. checkpoint-20/trainer_state.json +155 -0
  26. checkpoint-20/training_args.bin +3 -0
  27. checkpoint-30/README.md +21 -0
  28. checkpoint-30/adapter_config.json +26 -0
  29. checkpoint-30/adapter_model.bin +3 -0
  30. checkpoint-30/adapter_model/README.md +21 -0
  31. checkpoint-30/adapter_model/adapter_config.json +26 -0
  32. checkpoint-30/adapter_model/adapter_model.bin +3 -0
  33. checkpoint-30/optimizer.pt +3 -0
  34. checkpoint-30/rng_state.pth +3 -0
  35. checkpoint-30/scheduler.pt +3 -0
  36. checkpoint-30/trainer_state.json +223 -0
  37. checkpoint-30/training_args.bin +3 -0
  38. checkpoint-40/README.md +21 -0
  39. checkpoint-40/adapter_config.json +26 -0
  40. checkpoint-40/adapter_model.bin +3 -0
  41. checkpoint-40/adapter_model/README.md +21 -0
  42. checkpoint-40/adapter_model/adapter_config.json +26 -0
  43. checkpoint-40/adapter_model/adapter_model.bin +3 -0
  44. checkpoint-40/optimizer.pt +3 -0
  45. checkpoint-40/rng_state.pth +3 -0
  46. checkpoint-40/scheduler.pt +3 -0
  47. checkpoint-40/trainer_state.json +291 -0
  48. checkpoint-40/training_args.bin +3 -0
  49. checkpoint-50/README.md +21 -0
  50. checkpoint-50/adapter_config.json +26 -0
.ipynb_checkpoints/Untitled-checkpoint.ipynb ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "cells": [],
3
+ "metadata": {},
4
+ "nbformat": 4,
5
+ "nbformat_minor": 5
6
+ }
Untitled.ipynb ADDED
@@ -0,0 +1,50 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "cells": [
3
+ {
4
+ "cell_type": "code",
5
+ "execution_count": null,
6
+ "id": "127668ee-44f0-4438-9337-e7c4a486aea3",
7
+ "metadata": {},
8
+ "outputs": [],
9
+ "source": [
10
+ "from huggingface_hub import HfApi\n",
11
+ "\n",
12
+ "api = HfApi()\n",
13
+ "\n",
14
+ "# Upload all the content from the local folder to your remote Space.\n",
15
+ "# By default, files are uploaded at the root of the repo\n",
16
+ "\n",
17
+ "api.upload_folder(\n",
18
+ "\n",
19
+ " folder_path=r\"C:\\dataset\\New folder\",\n",
20
+ "\n",
21
+ " repo_id=\"MinervaAI/Random-roleplay-instruction\",\n",
22
+ "\n",
23
+ " repo_type=\"dataset\",\n",
24
+ "\n",
25
+ ")"
26
+ ]
27
+ }
28
+ ],
29
+ "metadata": {
30
+ "kernelspec": {
31
+ "display_name": "Python 3 (ipykernel)",
32
+ "language": "python",
33
+ "name": "python3"
34
+ },
35
+ "language_info": {
36
+ "codemirror_mode": {
37
+ "name": "ipython",
38
+ "version": 3
39
+ },
40
+ "file_extension": ".py",
41
+ "mimetype": "text/x-python",
42
+ "name": "python",
43
+ "nbconvert_exporter": "python",
44
+ "pygments_lexer": "ipython3",
45
+ "version": "3.10.12"
46
+ }
47
+ },
48
+ "nbformat": 4,
49
+ "nbformat_minor": 5
50
+ }
adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "NousResearch/Llama-2-13b-hf",
4
+ "bias": "none",
5
+ "fan_in_fan_out": null,
6
+ "inference_mode": false,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 8,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "v_proj",
19
+ "up_proj",
20
+ "gate_proj",
21
+ "o_proj",
22
+ "k_proj",
23
+ "down_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
added_tokens.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "<pad>": 32000
3
+ }
checkpoint-10/README.md ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: bitsandbytes
9
+ - load_in_8bit: False
10
+ - load_in_4bit: True
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: nf4
16
+ - bnb_4bit_use_double_quant: True
17
+ - bnb_4bit_compute_dtype: bfloat16
18
+
19
+ The following `bitsandbytes` quantization config was used during training:
20
+ - quant_method: bitsandbytes
21
+ - load_in_8bit: False
22
+ - load_in_4bit: True
23
+ - llm_int8_threshold: 6.0
24
+ - llm_int8_skip_modules: None
25
+ - llm_int8_enable_fp32_cpu_offload: False
26
+ - llm_int8_has_fp16_weight: False
27
+ - bnb_4bit_quant_type: nf4
28
+ - bnb_4bit_use_double_quant: True
29
+ - bnb_4bit_compute_dtype: bfloat16
30
+ ### Framework versions
31
+
32
+ - PEFT 0.6.0.dev0
33
+
34
+ - PEFT 0.6.0.dev0
checkpoint-10/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "NousResearch/Llama-2-13b-hf",
4
+ "bias": "none",
5
+ "fan_in_fan_out": null,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 8,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "v_proj",
19
+ "up_proj",
20
+ "gate_proj",
21
+ "o_proj",
22
+ "k_proj",
23
+ "down_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-10/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d73dd5f0590e7395a03feab2c12262e8144fdb10f84d6cc08f5c2b521c7d832
3
+ size 62788109
checkpoint-10/adapter_model/README.md ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: bitsandbytes
9
+ - load_in_8bit: False
10
+ - load_in_4bit: True
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: nf4
16
+ - bnb_4bit_use_double_quant: True
17
+ - bnb_4bit_compute_dtype: bfloat16
18
+
19
+ The following `bitsandbytes` quantization config was used during training:
20
+ - quant_method: bitsandbytes
21
+ - load_in_8bit: False
22
+ - load_in_4bit: True
23
+ - llm_int8_threshold: 6.0
24
+ - llm_int8_skip_modules: None
25
+ - llm_int8_enable_fp32_cpu_offload: False
26
+ - llm_int8_has_fp16_weight: False
27
+ - bnb_4bit_quant_type: nf4
28
+ - bnb_4bit_use_double_quant: True
29
+ - bnb_4bit_compute_dtype: bfloat16
30
+ ### Framework versions
31
+
32
+ - PEFT 0.6.0.dev0
33
+
34
+ - PEFT 0.6.0.dev0
checkpoint-10/adapter_model/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "NousResearch/Llama-2-13b-hf",
4
+ "bias": "none",
5
+ "fan_in_fan_out": null,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 8,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "v_proj",
19
+ "up_proj",
20
+ "gate_proj",
21
+ "o_proj",
22
+ "k_proj",
23
+ "down_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-10/adapter_model/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d73dd5f0590e7395a03feab2c12262e8144fdb10f84d6cc08f5c2b521c7d832
3
+ size 62788109
checkpoint-10/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:93fc15492ebf7acd0d7da2effdc336db5cf94da3d033e4bfcba017a5ba9ff578
3
+ size 250681597
checkpoint-10/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b8d0d39ee3f495298b1364d1a694b15161510d7f6b4a57e4ad295a4590655a19
3
+ size 14575
checkpoint-10/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e3f23bc239cbe3e7a4f862f2448a71aa49fc3b361c6c5866cc00242a779c0be4
3
+ size 627
checkpoint-10/trainer_state.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 2.098437547683716,
3
+ "best_model_checkpoint": "./qlora-out-kimiko-test2/checkpoint-10",
4
+ "epoch": 0.25848142164781907,
5
+ "eval_steps": 10,
6
+ "global_step": 10,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.03,
13
+ "learning_rate": 1e-05,
14
+ "loss": 1.5707,
15
+ "step": 1
16
+ },
17
+ {
18
+ "epoch": 0.05,
19
+ "learning_rate": 2e-05,
20
+ "loss": 1.5621,
21
+ "step": 2
22
+ },
23
+ {
24
+ "epoch": 0.08,
25
+ "learning_rate": 3e-05,
26
+ "loss": 1.4812,
27
+ "step": 3
28
+ },
29
+ {
30
+ "epoch": 0.1,
31
+ "learning_rate": 4e-05,
32
+ "loss": 1.5197,
33
+ "step": 4
34
+ },
35
+ {
36
+ "epoch": 0.13,
37
+ "learning_rate": 5e-05,
38
+ "loss": 1.5567,
39
+ "step": 5
40
+ },
41
+ {
42
+ "epoch": 0.16,
43
+ "learning_rate": 5e-05,
44
+ "loss": 1.4645,
45
+ "step": 6
46
+ },
47
+ {
48
+ "epoch": 0.18,
49
+ "learning_rate": 5e-05,
50
+ "loss": 1.6122,
51
+ "step": 7
52
+ },
53
+ {
54
+ "epoch": 0.21,
55
+ "learning_rate": 5e-05,
56
+ "loss": 1.5596,
57
+ "step": 8
58
+ },
59
+ {
60
+ "epoch": 0.23,
61
+ "learning_rate": 5e-05,
62
+ "loss": 1.5608,
63
+ "step": 9
64
+ },
65
+ {
66
+ "epoch": 0.26,
67
+ "learning_rate": 5e-05,
68
+ "loss": 1.5456,
69
+ "step": 10
70
+ },
71
+ {
72
+ "epoch": 0.26,
73
+ "eval_loss": 2.098437547683716,
74
+ "eval_runtime": 119.6161,
75
+ "eval_samples_per_second": 1.555,
76
+ "eval_steps_per_second": 0.777,
77
+ "step": 10
78
+ }
79
+ ],
80
+ "logging_steps": 1,
81
+ "max_steps": 114,
82
+ "num_train_epochs": 3,
83
+ "save_steps": 10,
84
+ "total_flos": 4.582587092041728e+16,
85
+ "trial_name": null,
86
+ "trial_params": null
87
+ }
checkpoint-10/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca362c6e6a1bbe2523b2190a501c92d6dbb3db6186bef551619d83852cca3df1
3
+ size 4219
checkpoint-20/README.md ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: bitsandbytes
9
+ - load_in_8bit: False
10
+ - load_in_4bit: True
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: nf4
16
+ - bnb_4bit_use_double_quant: True
17
+ - bnb_4bit_compute_dtype: bfloat16
18
+ ### Framework versions
19
+
20
+
21
+ - PEFT 0.6.0.dev0
checkpoint-20/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "NousResearch/Llama-2-13b-hf",
4
+ "bias": "none",
5
+ "fan_in_fan_out": null,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 8,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "v_proj",
19
+ "up_proj",
20
+ "gate_proj",
21
+ "o_proj",
22
+ "k_proj",
23
+ "down_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-20/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:03d0c38222a997a52bfe76857917321fb2bc8d10553b4da4ff8390643c25f962
3
+ size 62788109
checkpoint-20/adapter_model/README.md ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: bitsandbytes
9
+ - load_in_8bit: False
10
+ - load_in_4bit: True
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: nf4
16
+ - bnb_4bit_use_double_quant: True
17
+ - bnb_4bit_compute_dtype: bfloat16
18
+ ### Framework versions
19
+
20
+
21
+ - PEFT 0.6.0.dev0
checkpoint-20/adapter_model/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "NousResearch/Llama-2-13b-hf",
4
+ "bias": "none",
5
+ "fan_in_fan_out": null,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 8,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "v_proj",
19
+ "up_proj",
20
+ "gate_proj",
21
+ "o_proj",
22
+ "k_proj",
23
+ "down_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-20/adapter_model/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:03d0c38222a997a52bfe76857917321fb2bc8d10553b4da4ff8390643c25f962
3
+ size 62788109
checkpoint-20/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:56e8c6c5c1d0e07913575e95580988f8c4d4d451e12e22619c79c5df6b48dc53
3
+ size 250681597
checkpoint-20/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6cf66b0a499885f833314015600150cb6bfa74e1505e5608890c7c4ba655a6ba
3
+ size 14575
checkpoint-20/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e64bd2871a807ae2da11c5073d611b4d4223c336499ef21e9e856aaa448a1a35
3
+ size 627
checkpoint-20/trainer_state.json ADDED
@@ -0,0 +1,155 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 2.066981077194214,
3
+ "best_model_checkpoint": "./qlora-out-kimiko-test2/checkpoint-20",
4
+ "epoch": 0.5169628432956381,
5
+ "eval_steps": 10,
6
+ "global_step": 20,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.03,
13
+ "learning_rate": 1e-05,
14
+ "loss": 1.5707,
15
+ "step": 1
16
+ },
17
+ {
18
+ "epoch": 0.05,
19
+ "learning_rate": 2e-05,
20
+ "loss": 1.5621,
21
+ "step": 2
22
+ },
23
+ {
24
+ "epoch": 0.08,
25
+ "learning_rate": 3e-05,
26
+ "loss": 1.4812,
27
+ "step": 3
28
+ },
29
+ {
30
+ "epoch": 0.1,
31
+ "learning_rate": 4e-05,
32
+ "loss": 1.5197,
33
+ "step": 4
34
+ },
35
+ {
36
+ "epoch": 0.13,
37
+ "learning_rate": 5e-05,
38
+ "loss": 1.5567,
39
+ "step": 5
40
+ },
41
+ {
42
+ "epoch": 0.16,
43
+ "learning_rate": 5e-05,
44
+ "loss": 1.4645,
45
+ "step": 6
46
+ },
47
+ {
48
+ "epoch": 0.18,
49
+ "learning_rate": 5e-05,
50
+ "loss": 1.6122,
51
+ "step": 7
52
+ },
53
+ {
54
+ "epoch": 0.21,
55
+ "learning_rate": 5e-05,
56
+ "loss": 1.5596,
57
+ "step": 8
58
+ },
59
+ {
60
+ "epoch": 0.23,
61
+ "learning_rate": 5e-05,
62
+ "loss": 1.5608,
63
+ "step": 9
64
+ },
65
+ {
66
+ "epoch": 0.26,
67
+ "learning_rate": 5e-05,
68
+ "loss": 1.5456,
69
+ "step": 10
70
+ },
71
+ {
72
+ "epoch": 0.26,
73
+ "eval_loss": 2.098437547683716,
74
+ "eval_runtime": 119.6161,
75
+ "eval_samples_per_second": 1.555,
76
+ "eval_steps_per_second": 0.777,
77
+ "step": 10
78
+ },
79
+ {
80
+ "epoch": 0.28,
81
+ "learning_rate": 5e-05,
82
+ "loss": 1.5645,
83
+ "step": 11
84
+ },
85
+ {
86
+ "epoch": 0.31,
87
+ "learning_rate": 5e-05,
88
+ "loss": 1.538,
89
+ "step": 12
90
+ },
91
+ {
92
+ "epoch": 0.34,
93
+ "learning_rate": 5e-05,
94
+ "loss": 1.6388,
95
+ "step": 13
96
+ },
97
+ {
98
+ "epoch": 0.36,
99
+ "learning_rate": 5e-05,
100
+ "loss": 1.4943,
101
+ "step": 14
102
+ },
103
+ {
104
+ "epoch": 0.39,
105
+ "learning_rate": 5e-05,
106
+ "loss": 1.5469,
107
+ "step": 15
108
+ },
109
+ {
110
+ "epoch": 0.41,
111
+ "learning_rate": 5e-05,
112
+ "loss": 1.6149,
113
+ "step": 16
114
+ },
115
+ {
116
+ "epoch": 0.44,
117
+ "learning_rate": 5e-05,
118
+ "loss": 1.5345,
119
+ "step": 17
120
+ },
121
+ {
122
+ "epoch": 0.47,
123
+ "learning_rate": 5e-05,
124
+ "loss": 1.4903,
125
+ "step": 18
126
+ },
127
+ {
128
+ "epoch": 0.49,
129
+ "learning_rate": 5e-05,
130
+ "loss": 1.5499,
131
+ "step": 19
132
+ },
133
+ {
134
+ "epoch": 0.52,
135
+ "learning_rate": 5e-05,
136
+ "loss": 1.5934,
137
+ "step": 20
138
+ },
139
+ {
140
+ "epoch": 0.52,
141
+ "eval_loss": 2.066981077194214,
142
+ "eval_runtime": 119.5781,
143
+ "eval_samples_per_second": 1.555,
144
+ "eval_steps_per_second": 0.778,
145
+ "step": 20
146
+ }
147
+ ],
148
+ "logging_steps": 1,
149
+ "max_steps": 114,
150
+ "num_train_epochs": 3,
151
+ "save_steps": 10,
152
+ "total_flos": 9.195307914756096e+16,
153
+ "trial_name": null,
154
+ "trial_params": null
155
+ }
checkpoint-20/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca362c6e6a1bbe2523b2190a501c92d6dbb3db6186bef551619d83852cca3df1
3
+ size 4219
checkpoint-30/README.md ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: bitsandbytes
9
+ - load_in_8bit: False
10
+ - load_in_4bit: True
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: nf4
16
+ - bnb_4bit_use_double_quant: True
17
+ - bnb_4bit_compute_dtype: bfloat16
18
+ ### Framework versions
19
+
20
+
21
+ - PEFT 0.6.0.dev0
checkpoint-30/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "NousResearch/Llama-2-13b-hf",
4
+ "bias": "none",
5
+ "fan_in_fan_out": null,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 8,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "v_proj",
19
+ "up_proj",
20
+ "gate_proj",
21
+ "o_proj",
22
+ "k_proj",
23
+ "down_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-30/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4276aa46f33d393edc2d19308e186a1f07580d76eddb0bb6339bc4f3e80b9f58
3
+ size 62788109
checkpoint-30/adapter_model/README.md ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: bitsandbytes
9
+ - load_in_8bit: False
10
+ - load_in_4bit: True
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: nf4
16
+ - bnb_4bit_use_double_quant: True
17
+ - bnb_4bit_compute_dtype: bfloat16
18
+ ### Framework versions
19
+
20
+
21
+ - PEFT 0.6.0.dev0
checkpoint-30/adapter_model/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "NousResearch/Llama-2-13b-hf",
4
+ "bias": "none",
5
+ "fan_in_fan_out": null,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 8,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "v_proj",
19
+ "up_proj",
20
+ "gate_proj",
21
+ "o_proj",
22
+ "k_proj",
23
+ "down_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-30/adapter_model/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4276aa46f33d393edc2d19308e186a1f07580d76eddb0bb6339bc4f3e80b9f58
3
+ size 62788109
checkpoint-30/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b18fc6beca12122a3d4309af96449623e23a0d54f4b1c5088810cb8933a25809
3
+ size 250681597
checkpoint-30/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cbf64f9f720011db899c09efa51061d9067d83e6fc2235e5bb9d7087d72402fd
3
+ size 14575
checkpoint-30/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:298c19c2d0215ea63da7419132c9dac6c9c75fdd531e32dfc1cb87dbb0aa8259
3
+ size 627
checkpoint-30/trainer_state.json ADDED
@@ -0,0 +1,223 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 2.046339750289917,
3
+ "best_model_checkpoint": "./qlora-out-kimiko-test2/checkpoint-30",
4
+ "epoch": 0.7754442649434572,
5
+ "eval_steps": 10,
6
+ "global_step": 30,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.03,
13
+ "learning_rate": 1e-05,
14
+ "loss": 1.5707,
15
+ "step": 1
16
+ },
17
+ {
18
+ "epoch": 0.05,
19
+ "learning_rate": 2e-05,
20
+ "loss": 1.5621,
21
+ "step": 2
22
+ },
23
+ {
24
+ "epoch": 0.08,
25
+ "learning_rate": 3e-05,
26
+ "loss": 1.4812,
27
+ "step": 3
28
+ },
29
+ {
30
+ "epoch": 0.1,
31
+ "learning_rate": 4e-05,
32
+ "loss": 1.5197,
33
+ "step": 4
34
+ },
35
+ {
36
+ "epoch": 0.13,
37
+ "learning_rate": 5e-05,
38
+ "loss": 1.5567,
39
+ "step": 5
40
+ },
41
+ {
42
+ "epoch": 0.16,
43
+ "learning_rate": 5e-05,
44
+ "loss": 1.4645,
45
+ "step": 6
46
+ },
47
+ {
48
+ "epoch": 0.18,
49
+ "learning_rate": 5e-05,
50
+ "loss": 1.6122,
51
+ "step": 7
52
+ },
53
+ {
54
+ "epoch": 0.21,
55
+ "learning_rate": 5e-05,
56
+ "loss": 1.5596,
57
+ "step": 8
58
+ },
59
+ {
60
+ "epoch": 0.23,
61
+ "learning_rate": 5e-05,
62
+ "loss": 1.5608,
63
+ "step": 9
64
+ },
65
+ {
66
+ "epoch": 0.26,
67
+ "learning_rate": 5e-05,
68
+ "loss": 1.5456,
69
+ "step": 10
70
+ },
71
+ {
72
+ "epoch": 0.26,
73
+ "eval_loss": 2.098437547683716,
74
+ "eval_runtime": 119.6161,
75
+ "eval_samples_per_second": 1.555,
76
+ "eval_steps_per_second": 0.777,
77
+ "step": 10
78
+ },
79
+ {
80
+ "epoch": 0.28,
81
+ "learning_rate": 5e-05,
82
+ "loss": 1.5645,
83
+ "step": 11
84
+ },
85
+ {
86
+ "epoch": 0.31,
87
+ "learning_rate": 5e-05,
88
+ "loss": 1.538,
89
+ "step": 12
90
+ },
91
+ {
92
+ "epoch": 0.34,
93
+ "learning_rate": 5e-05,
94
+ "loss": 1.6388,
95
+ "step": 13
96
+ },
97
+ {
98
+ "epoch": 0.36,
99
+ "learning_rate": 5e-05,
100
+ "loss": 1.4943,
101
+ "step": 14
102
+ },
103
+ {
104
+ "epoch": 0.39,
105
+ "learning_rate": 5e-05,
106
+ "loss": 1.5469,
107
+ "step": 15
108
+ },
109
+ {
110
+ "epoch": 0.41,
111
+ "learning_rate": 5e-05,
112
+ "loss": 1.6149,
113
+ "step": 16
114
+ },
115
+ {
116
+ "epoch": 0.44,
117
+ "learning_rate": 5e-05,
118
+ "loss": 1.5345,
119
+ "step": 17
120
+ },
121
+ {
122
+ "epoch": 0.47,
123
+ "learning_rate": 5e-05,
124
+ "loss": 1.4903,
125
+ "step": 18
126
+ },
127
+ {
128
+ "epoch": 0.49,
129
+ "learning_rate": 5e-05,
130
+ "loss": 1.5499,
131
+ "step": 19
132
+ },
133
+ {
134
+ "epoch": 0.52,
135
+ "learning_rate": 5e-05,
136
+ "loss": 1.5934,
137
+ "step": 20
138
+ },
139
+ {
140
+ "epoch": 0.52,
141
+ "eval_loss": 2.066981077194214,
142
+ "eval_runtime": 119.5781,
143
+ "eval_samples_per_second": 1.555,
144
+ "eval_steps_per_second": 0.778,
145
+ "step": 20
146
+ },
147
+ {
148
+ "epoch": 0.54,
149
+ "learning_rate": 5e-05,
150
+ "loss": 1.4554,
151
+ "step": 21
152
+ },
153
+ {
154
+ "epoch": 0.57,
155
+ "learning_rate": 5e-05,
156
+ "loss": 1.5512,
157
+ "step": 22
158
+ },
159
+ {
160
+ "epoch": 0.59,
161
+ "learning_rate": 5e-05,
162
+ "loss": 1.4636,
163
+ "step": 23
164
+ },
165
+ {
166
+ "epoch": 0.62,
167
+ "learning_rate": 5e-05,
168
+ "loss": 1.5398,
169
+ "step": 24
170
+ },
171
+ {
172
+ "epoch": 0.65,
173
+ "learning_rate": 5e-05,
174
+ "loss": 1.5623,
175
+ "step": 25
176
+ },
177
+ {
178
+ "epoch": 0.67,
179
+ "learning_rate": 5e-05,
180
+ "loss": 1.4658,
181
+ "step": 26
182
+ },
183
+ {
184
+ "epoch": 0.7,
185
+ "learning_rate": 5e-05,
186
+ "loss": 1.4723,
187
+ "step": 27
188
+ },
189
+ {
190
+ "epoch": 0.72,
191
+ "learning_rate": 5e-05,
192
+ "loss": 1.432,
193
+ "step": 28
194
+ },
195
+ {
196
+ "epoch": 0.75,
197
+ "learning_rate": 5e-05,
198
+ "loss": 1.4814,
199
+ "step": 29
200
+ },
201
+ {
202
+ "epoch": 0.78,
203
+ "learning_rate": 5e-05,
204
+ "loss": 1.4924,
205
+ "step": 30
206
+ },
207
+ {
208
+ "epoch": 0.78,
209
+ "eval_loss": 2.046339750289917,
210
+ "eval_runtime": 119.5771,
211
+ "eval_samples_per_second": 1.555,
212
+ "eval_steps_per_second": 0.778,
213
+ "step": 30
214
+ }
215
+ ],
216
+ "logging_steps": 1,
217
+ "max_steps": 114,
218
+ "num_train_epochs": 3,
219
+ "save_steps": 10,
220
+ "total_flos": 1.3777643892375552e+17,
221
+ "trial_name": null,
222
+ "trial_params": null
223
+ }
checkpoint-30/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca362c6e6a1bbe2523b2190a501c92d6dbb3db6186bef551619d83852cca3df1
3
+ size 4219
checkpoint-40/README.md ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: bitsandbytes
9
+ - load_in_8bit: False
10
+ - load_in_4bit: True
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: nf4
16
+ - bnb_4bit_use_double_quant: True
17
+ - bnb_4bit_compute_dtype: bfloat16
18
+ ### Framework versions
19
+
20
+
21
+ - PEFT 0.6.0.dev0
checkpoint-40/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "NousResearch/Llama-2-13b-hf",
4
+ "bias": "none",
5
+ "fan_in_fan_out": null,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 8,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "v_proj",
19
+ "up_proj",
20
+ "gate_proj",
21
+ "o_proj",
22
+ "k_proj",
23
+ "down_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-40/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:12f5a629e5ff4db207707d71a1c324a97283a0a371cbd435fb090ea711a9e21c
3
+ size 62788109
checkpoint-40/adapter_model/README.md ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: bitsandbytes
9
+ - load_in_8bit: False
10
+ - load_in_4bit: True
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: nf4
16
+ - bnb_4bit_use_double_quant: True
17
+ - bnb_4bit_compute_dtype: bfloat16
18
+ ### Framework versions
19
+
20
+
21
+ - PEFT 0.6.0.dev0
checkpoint-40/adapter_model/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "NousResearch/Llama-2-13b-hf",
4
+ "bias": "none",
5
+ "fan_in_fan_out": null,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 8,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "v_proj",
19
+ "up_proj",
20
+ "gate_proj",
21
+ "o_proj",
22
+ "k_proj",
23
+ "down_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-40/adapter_model/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:12f5a629e5ff4db207707d71a1c324a97283a0a371cbd435fb090ea711a9e21c
3
+ size 62788109
checkpoint-40/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d8c0a407654387b576e7ec5d641bcd7961201b07a96d39e22e3e9d77e541a513
3
+ size 250681597
checkpoint-40/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6ef3dbc703c54e9cba3cb628e0e6ae4c7f37310ab3398f30c0c5503dcb749f77
3
+ size 14575
checkpoint-40/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9d048c63c660c9f2cb6345c4f756c8abee0e570482b06bf9c08827d79b257394
3
+ size 627
checkpoint-40/trainer_state.json ADDED
@@ -0,0 +1,291 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 2.0354697704315186,
3
+ "best_model_checkpoint": "./qlora-out-kimiko-test2/checkpoint-40",
4
+ "epoch": 1.0339256865912763,
5
+ "eval_steps": 10,
6
+ "global_step": 40,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.03,
13
+ "learning_rate": 1e-05,
14
+ "loss": 1.5707,
15
+ "step": 1
16
+ },
17
+ {
18
+ "epoch": 0.05,
19
+ "learning_rate": 2e-05,
20
+ "loss": 1.5621,
21
+ "step": 2
22
+ },
23
+ {
24
+ "epoch": 0.08,
25
+ "learning_rate": 3e-05,
26
+ "loss": 1.4812,
27
+ "step": 3
28
+ },
29
+ {
30
+ "epoch": 0.1,
31
+ "learning_rate": 4e-05,
32
+ "loss": 1.5197,
33
+ "step": 4
34
+ },
35
+ {
36
+ "epoch": 0.13,
37
+ "learning_rate": 5e-05,
38
+ "loss": 1.5567,
39
+ "step": 5
40
+ },
41
+ {
42
+ "epoch": 0.16,
43
+ "learning_rate": 5e-05,
44
+ "loss": 1.4645,
45
+ "step": 6
46
+ },
47
+ {
48
+ "epoch": 0.18,
49
+ "learning_rate": 5e-05,
50
+ "loss": 1.6122,
51
+ "step": 7
52
+ },
53
+ {
54
+ "epoch": 0.21,
55
+ "learning_rate": 5e-05,
56
+ "loss": 1.5596,
57
+ "step": 8
58
+ },
59
+ {
60
+ "epoch": 0.23,
61
+ "learning_rate": 5e-05,
62
+ "loss": 1.5608,
63
+ "step": 9
64
+ },
65
+ {
66
+ "epoch": 0.26,
67
+ "learning_rate": 5e-05,
68
+ "loss": 1.5456,
69
+ "step": 10
70
+ },
71
+ {
72
+ "epoch": 0.26,
73
+ "eval_loss": 2.098437547683716,
74
+ "eval_runtime": 119.6161,
75
+ "eval_samples_per_second": 1.555,
76
+ "eval_steps_per_second": 0.777,
77
+ "step": 10
78
+ },
79
+ {
80
+ "epoch": 0.28,
81
+ "learning_rate": 5e-05,
82
+ "loss": 1.5645,
83
+ "step": 11
84
+ },
85
+ {
86
+ "epoch": 0.31,
87
+ "learning_rate": 5e-05,
88
+ "loss": 1.538,
89
+ "step": 12
90
+ },
91
+ {
92
+ "epoch": 0.34,
93
+ "learning_rate": 5e-05,
94
+ "loss": 1.6388,
95
+ "step": 13
96
+ },
97
+ {
98
+ "epoch": 0.36,
99
+ "learning_rate": 5e-05,
100
+ "loss": 1.4943,
101
+ "step": 14
102
+ },
103
+ {
104
+ "epoch": 0.39,
105
+ "learning_rate": 5e-05,
106
+ "loss": 1.5469,
107
+ "step": 15
108
+ },
109
+ {
110
+ "epoch": 0.41,
111
+ "learning_rate": 5e-05,
112
+ "loss": 1.6149,
113
+ "step": 16
114
+ },
115
+ {
116
+ "epoch": 0.44,
117
+ "learning_rate": 5e-05,
118
+ "loss": 1.5345,
119
+ "step": 17
120
+ },
121
+ {
122
+ "epoch": 0.47,
123
+ "learning_rate": 5e-05,
124
+ "loss": 1.4903,
125
+ "step": 18
126
+ },
127
+ {
128
+ "epoch": 0.49,
129
+ "learning_rate": 5e-05,
130
+ "loss": 1.5499,
131
+ "step": 19
132
+ },
133
+ {
134
+ "epoch": 0.52,
135
+ "learning_rate": 5e-05,
136
+ "loss": 1.5934,
137
+ "step": 20
138
+ },
139
+ {
140
+ "epoch": 0.52,
141
+ "eval_loss": 2.066981077194214,
142
+ "eval_runtime": 119.5781,
143
+ "eval_samples_per_second": 1.555,
144
+ "eval_steps_per_second": 0.778,
145
+ "step": 20
146
+ },
147
+ {
148
+ "epoch": 0.54,
149
+ "learning_rate": 5e-05,
150
+ "loss": 1.4554,
151
+ "step": 21
152
+ },
153
+ {
154
+ "epoch": 0.57,
155
+ "learning_rate": 5e-05,
156
+ "loss": 1.5512,
157
+ "step": 22
158
+ },
159
+ {
160
+ "epoch": 0.59,
161
+ "learning_rate": 5e-05,
162
+ "loss": 1.4636,
163
+ "step": 23
164
+ },
165
+ {
166
+ "epoch": 0.62,
167
+ "learning_rate": 5e-05,
168
+ "loss": 1.5398,
169
+ "step": 24
170
+ },
171
+ {
172
+ "epoch": 0.65,
173
+ "learning_rate": 5e-05,
174
+ "loss": 1.5623,
175
+ "step": 25
176
+ },
177
+ {
178
+ "epoch": 0.67,
179
+ "learning_rate": 5e-05,
180
+ "loss": 1.4658,
181
+ "step": 26
182
+ },
183
+ {
184
+ "epoch": 0.7,
185
+ "learning_rate": 5e-05,
186
+ "loss": 1.4723,
187
+ "step": 27
188
+ },
189
+ {
190
+ "epoch": 0.72,
191
+ "learning_rate": 5e-05,
192
+ "loss": 1.432,
193
+ "step": 28
194
+ },
195
+ {
196
+ "epoch": 0.75,
197
+ "learning_rate": 5e-05,
198
+ "loss": 1.4814,
199
+ "step": 29
200
+ },
201
+ {
202
+ "epoch": 0.78,
203
+ "learning_rate": 5e-05,
204
+ "loss": 1.4924,
205
+ "step": 30
206
+ },
207
+ {
208
+ "epoch": 0.78,
209
+ "eval_loss": 2.046339750289917,
210
+ "eval_runtime": 119.5771,
211
+ "eval_samples_per_second": 1.555,
212
+ "eval_steps_per_second": 0.778,
213
+ "step": 30
214
+ },
215
+ {
216
+ "epoch": 0.8,
217
+ "learning_rate": 5e-05,
218
+ "loss": 1.5809,
219
+ "step": 31
220
+ },
221
+ {
222
+ "epoch": 0.83,
223
+ "learning_rate": 5e-05,
224
+ "loss": 1.4803,
225
+ "step": 32
226
+ },
227
+ {
228
+ "epoch": 0.85,
229
+ "learning_rate": 5e-05,
230
+ "loss": 1.4878,
231
+ "step": 33
232
+ },
233
+ {
234
+ "epoch": 0.88,
235
+ "learning_rate": 5e-05,
236
+ "loss": 1.3871,
237
+ "step": 34
238
+ },
239
+ {
240
+ "epoch": 0.9,
241
+ "learning_rate": 5e-05,
242
+ "loss": 1.5151,
243
+ "step": 35
244
+ },
245
+ {
246
+ "epoch": 0.93,
247
+ "learning_rate": 5e-05,
248
+ "loss": 1.4212,
249
+ "step": 36
250
+ },
251
+ {
252
+ "epoch": 0.96,
253
+ "learning_rate": 5e-05,
254
+ "loss": 1.6284,
255
+ "step": 37
256
+ },
257
+ {
258
+ "epoch": 0.98,
259
+ "learning_rate": 5e-05,
260
+ "loss": 1.5002,
261
+ "step": 38
262
+ },
263
+ {
264
+ "epoch": 1.01,
265
+ "learning_rate": 5e-05,
266
+ "loss": 1.4452,
267
+ "step": 39
268
+ },
269
+ {
270
+ "epoch": 1.03,
271
+ "learning_rate": 5e-05,
272
+ "loss": 1.4399,
273
+ "step": 40
274
+ },
275
+ {
276
+ "epoch": 1.03,
277
+ "eval_loss": 2.0354697704315186,
278
+ "eval_runtime": 119.5875,
279
+ "eval_samples_per_second": 1.555,
280
+ "eval_steps_per_second": 0.778,
281
+ "step": 40
282
+ }
283
+ ],
284
+ "logging_steps": 1,
285
+ "max_steps": 114,
286
+ "num_train_epochs": 3,
287
+ "save_steps": 10,
288
+ "total_flos": 1.8296447921160192e+17,
289
+ "trial_name": null,
290
+ "trial_params": null
291
+ }
checkpoint-40/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca362c6e6a1bbe2523b2190a501c92d6dbb3db6186bef551619d83852cca3df1
3
+ size 4219
checkpoint-50/README.md ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: bitsandbytes
9
+ - load_in_8bit: False
10
+ - load_in_4bit: True
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: nf4
16
+ - bnb_4bit_use_double_quant: True
17
+ - bnb_4bit_compute_dtype: bfloat16
18
+ ### Framework versions
19
+
20
+
21
+ - PEFT 0.6.0.dev0
checkpoint-50/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "NousResearch/Llama-2-13b-hf",
4
+ "bias": "none",
5
+ "fan_in_fan_out": null,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 8,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "v_proj",
19
+ "up_proj",
20
+ "gate_proj",
21
+ "o_proj",
22
+ "k_proj",
23
+ "down_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }