Upload folder using huggingface_hub

Qwen/Qwen1.5_1.8B_ledgar/README.md

@@ -0,0 +1,93 @@
+---
+license: other
+base_model: Qwen/Qwen1.5-1.8B
+tags:
+- generated_from_trainer
+metrics:
+- accuracy
+model-index:
+- name: Qwen1.5_1.8B_ledgar
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# Qwen1.5_1.8B_ledgar
+
+This model is a fine-tuned version of [Qwen/Qwen1.5-1.8B](https://huggingface.co/Qwen/Qwen1.5-1.8B) on an unknown dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.5064
+- Accuracy: 0.8669
+- F1 Macro: 0.7902
+- F1 Micro: 0.8669
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 5e-06
+- train_batch_size: 32
+- eval_batch_size: 32
+- seed: 42
+- distributed_type: multi-GPU
+- num_devices: 2
+- total_train_batch_size: 64
+- total_eval_batch_size: 64
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: linear
+- num_epochs: 3.0
+
+### Training results
+
+| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 Macro | F1 Micro |
+|:-------------:|:-----:|:----:|:---------------:|:--------:|:--------:|:--------:|
+| 1.3077        | 0.11  | 100  | 1.0945          | 0.7277   | 0.5771   | 0.7277   |
+| 0.8627        | 0.21  | 200  | 0.8368          | 0.7907   | 0.6657   | 0.7907   |
+| 0.7179        | 0.32  | 300  | 0.7824          | 0.7971   | 0.6862   | 0.7971   |
+| 0.6961        | 0.43  | 400  | 0.6952          | 0.8138   | 0.6992   | 0.8138   |
+| 0.745         | 0.53  | 500  | 0.6719          | 0.8121   | 0.7034   | 0.8121   |
+| 0.6505        | 0.64  | 600  | 0.6220          | 0.834    | 0.7469   | 0.834    |
+| 0.5914        | 0.75  | 700  | 0.6110          | 0.8362   | 0.7411   | 0.8362   |
+| 0.5837        | 0.85  | 800  | 0.5767          | 0.8385   | 0.7413   | 0.8385   |
+| 0.5218        | 0.96  | 900  | 0.5365          | 0.849    | 0.7703   | 0.849    |
+| 0.2632        | 1.07  | 1000 | 0.5504          | 0.8562   | 0.7684   | 0.8562   |
+| 0.2607        | 1.17  | 1100 | 0.5497          | 0.8525   | 0.7657   | 0.8525   |
+| 0.274         | 1.28  | 1200 | 0.5439          | 0.8584   | 0.7746   | 0.8584   |
+| 0.2216        | 1.39  | 1300 | 0.5687          | 0.8563   | 0.7754   | 0.8563   |
+| 0.2044        | 1.49  | 1400 | 0.5385          | 0.861    | 0.7820   | 0.861    |
+| 0.2508        | 1.6   | 1500 | 0.5658          | 0.8577   | 0.7711   | 0.8577   |
+| 0.2513        | 1.71  | 1600 | 0.5367          | 0.8589   | 0.7872   | 0.8589   |
+| 0.2787        | 1.81  | 1700 | 0.5133          | 0.8653   | 0.7903   | 0.8653   |
+| 0.2357        | 1.92  | 1800 | 0.5064          | 0.8669   | 0.7902   | 0.8669   |
+| 0.049         | 2.03  | 1900 | 0.5344          | 0.8719   | 0.7978   | 0.8719   |
+| 0.0298        | 2.13  | 2000 | 0.5762          | 0.8737   | 0.7992   | 0.8737   |
+| 0.0427        | 2.24  | 2100 | 0.5961          | 0.8708   | 0.7976   | 0.8708   |
+| 0.036         | 2.35  | 2200 | 0.6128          | 0.8728   | 0.7988   | 0.8728   |
+| 0.0551        | 2.45  | 2300 | 0.6165          | 0.8708   | 0.7976   | 0.8708   |
+| 0.0392        | 2.56  | 2400 | 0.6023          | 0.8749   | 0.8038   | 0.8749   |
+| 0.0364        | 2.67  | 2500 | 0.6168          | 0.8729   | 0.8001   | 0.8729   |
+| 0.0416        | 2.77  | 2600 | 0.6103          | 0.8753   | 0.8048   | 0.8753   |
+| 0.0353        | 2.88  | 2700 | 0.6118          | 0.8749   | 0.8054   | 0.8749   |
+| 0.0308        | 2.99  | 2800 | 0.6114          | 0.875    | 0.8057   | 0.875    |
+
+
+### Framework versions
+
+- Transformers 4.39.0.dev0
+- Pytorch 2.2.1+cu121
+- Datasets 2.18.0
+- Tokenizers 0.15.2

Qwen/Qwen1.5_1.8B_ledgar/added_tokens.json

@@ -0,0 +1,5 @@
+{
+  "<|endoftext|>": 151643,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644
+}

Qwen/Qwen1.5_1.8B_ledgar/all_results.json

@@ -0,0 +1,23 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8669,
+    "eval_f1_macro": 0.7902403947168268,
+    "eval_f1_micro": 0.8669,
+    "eval_loss": 0.5063937306404114,
+    "eval_runtime": 24.4305,
+    "eval_samples": 10000,
+    "eval_samples_per_second": 409.324,
+    "eval_steps_per_second": 6.426,
+    "test_accuracy": 0.8664,
+    "test_f1_macro": 0.7974226514742132,
+    "test_f1_micro": 0.8664,
+    "test_loss": 0.532435953617096,
+    "test_runtime": 25.4113,
+    "test_samples_per_second": 393.525,
+    "test_steps_per_second": 6.178,
+    "train_loss": 0.42635450247932005,
+    "train_runtime": 3623.2348,
+    "train_samples": 60000,
+    "train_samples_per_second": 49.679,
+    "train_steps_per_second": 0.777
+}

Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800/added_tokens.json

@@ -0,0 +1,5 @@
+{
+  "<|endoftext|>": 151643,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644
+}

Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800/config.json

@@ -0,0 +1,235 @@
+{
+  "_name_or_path": "Qwen/Qwen1.5-1.8B",
+  "architectures": [
+    "Qwen2ForSequenceClassification"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "finetuning_task": "text-classification",
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "id2label": {
+    "0": "0",
+    "1": "1",
+    "2": "10",
+    "3": "11",
+    "4": "12",
+    "5": "13",
+    "6": "14",
+    "7": "15",
+    "8": "16",
+    "9": "17",
+    "10": "18",
+    "11": "19",
+    "12": "2",
+    "13": "20",
+    "14": "21",
+    "15": "22",
+    "16": "23",
+    "17": "24",
+    "18": "25",
+    "19": "26",
+    "20": "27",
+    "21": "28",
+    "22": "29",
+    "23": "3",
+    "24": "30",
+    "25": "31",
+    "26": "32",
+    "27": "33",
+    "28": "34",
+    "29": "35",
+    "30": "36",
+    "31": "37",
+    "32": "38",
+    "33": "39",
+    "34": "4",
+    "35": "40",
+    "36": "41",
+    "37": "42",
+    "38": "43",
+    "39": "44",
+    "40": "45",
+    "41": "46",
+    "42": "47",
+    "43": "48",
+    "44": "49",
+    "45": "5",
+    "46": "50",
+    "47": "51",
+    "48": "52",
+    "49": "53",
+    "50": "54",
+    "51": "55",
+    "52": "56",
+    "53": "57",
+    "54": "58",
+    "55": "59",
+    "56": "6",
+    "57": "60",
+    "58": "61",
+    "59": "62",
+    "60": "63",
+    "61": "64",
+    "62": "65",
+    "63": "66",
+    "64": "67",
+    "65": "68",
+    "66": "69",
+    "67": "7",
+    "68": "70",
+    "69": "71",
+    "70": "72",
+    "71": "73",
+    "72": "74",
+    "73": "75",
+    "74": "76",
+    "75": "77",
+    "76": "78",
+    "77": "79",
+    "78": "8",
+    "79": "80",
+    "80": "81",
+    "81": "82",
+    "82": "83",
+    "83": "84",
+    "84": "85",
+    "85": "86",
+    "86": "87",
+    "87": "88",
+    "88": "89",
+    "89": "9",
+    "90": "90",
+    "91": "91",
+    "92": "92",
+    "93": "93",
+    "94": "94",
+    "95": "95",
+    "96": "96",
+    "97": "97",
+    "98": "98",
+    "99": "99"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 5504,
+  "label2id": {
+    "0": 0,
+    "1": 1,
+    "10": 2,
+    "11": 3,
+    "12": 4,
+    "13": 5,
+    "14": 6,
+    "15": 7,
+    "16": 8,
+    "17": 9,
+    "18": 10,
+    "19": 11,
+    "2": 12,
+    "20": 13,
+    "21": 14,
+    "22": 15,
+    "23": 16,
+    "24": 17,
+    "25": 18,
+    "26": 19,
+    "27": 20,
+    "28": 21,
+    "29": 22,
+    "3": 23,
+    "30": 24,
+    "31": 25,
+    "32": 26,
+    "33": 27,
+    "34": 28,
+    "35": 29,
+    "36": 30,
+    "37": 31,
+    "38": 32,
+    "39": 33,
+    "4": 34,
+    "40": 35,
+    "41": 36,
+    "42": 37,
+    "43": 38,
+    "44": 39,
+    "45": 40,
+    "46": 41,
+    "47": 42,
+    "48": 43,
+    "49": 44,
+    "5": 45,
+    "50": 46,
+    "51": 47,
+    "52": 48,
+    "53": 49,
+    "54": 50,
+    "55": 51,
+    "56": 52,
+    "57": 53,
+    "58": 54,
+    "59": 55,
+    "6": 56,
+    "60": 57,
+    "61": 58,
+    "62": 59,
+    "63": 60,
+    "64": 61,
+    "65": 62,
+    "66": 63,
+    "67": 64,
+    "68": 65,
+    "69": 66,
+    "7": 67,
+    "70": 68,
+    "71": 69,
+    "72": 70,
+    "73": 71,
+    "74": 72,
+    "75": 73,
+    "76": 74,
+    "77": 75,
+    "78": 76,
+    "79": 77,
+    "8": 78,
+    "80": 79,
+    "81": 80,
+    "82": 81,
+    "83": 82,
+    "84": 83,
+    "85": 84,
+    "86": 85,
+    "87": 86,
+    "88": 87,
+    "89": 88,
+    "9": 89,
+    "90": 90,
+    "91": 91,
+    "92": 92,
+    "93": 93,
+    "94": 94,
+    "95": 95,
+    "96": 96,
+    "97": 97,
+    "98": 98,
+    "99": 99
+  },
+  "max_position_embeddings": 32768,
+  "max_window_layers": 21,
+  "model_type": "qwen2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 24,
+  "num_key_value_heads": 16,
+  "pad_token_id": 151643,
+  "problem_type": "single_label_classification",
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 1000000.0,
+  "sliding_window": 32768,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.39.0.dev0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151646
+}

Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800/global_step1800/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef1b558ae70d0d44599d87ecd1dfb077c81a23f338132f079e36f7b5290e7c08
+size 9151661452

Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800/global_step1800/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:370b65eed061f4813c54cd3f474407c386b95d6eb5825fc476c6a954ed387a5b
+size 9151666316

Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800/global_step1800/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:122da52fa34439283e1b14257cdd1cdeb266e23b92ffc56b8f05f799ec83fc3e
+size 3050629496

Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800/latest

@@ -0,0 +1 @@
+global_step1800

Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6be1da9a3d742bcd4f56af0dd627ec82b2b529a0f3a3259561a3da60f4f2dffc
+size 3050582504

Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:65f5ad184cd93fcb3486b7d9a9620ac3dd75a9ac32473b69036d8a9ea5d74a35
+size 14512

Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca245034f0fb0f22361e0627eba02734337b9a09a6d7dfcd1d3684fc6f29790d
+size 14512

Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73fb8dcc2e911a74681d6256ed942bb17afce93f5ee2a12d46277718ba595f5e
+size 1064

Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800/special_tokens_map.json

@@ -0,0 +1,14 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>"
+  ],
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<|endoftext|>"
+}

Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800/tokenizer_config.json

@@ -0,0 +1,43 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{% for message in messages %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "model_max_length": 32768,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800/trainer_state.json

@@ -0,0 +1,723 @@
+{
+  "best_metric": 0.5063937306404114,
+  "best_model_checkpoint": "../../experiments_checkpoints/MAdAiLab/Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800",
+  "epoch": 1.9189765458422174,
+  "eval_steps": 100,
+  "global_step": 1800,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.03,
+      "grad_norm": 80.4836196899414,
+      "learning_rate": 4.9555792466240235e-06,
+      "loss": 7.9887,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 60.927364349365234,
+      "learning_rate": 4.911158493248046e-06,
+      "loss": 3.1908,
+      "step": 50
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 49.68091583251953,
+      "learning_rate": 4.866737739872069e-06,
+      "loss": 1.7183,
+      "step": 75
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 55.176666259765625,
+      "learning_rate": 4.822316986496091e-06,
+      "loss": 1.3077,
+      "step": 100
+    },
+    {
+      "epoch": 0.11,
+      "eval_accuracy": 0.7277,
+      "eval_f1_macro": 0.5770831474844406,
+      "eval_f1_micro": 0.7277,
+      "eval_loss": 1.0944937467575073,
+      "eval_runtime": 25.447,
+      "eval_samples_per_second": 392.973,
+      "eval_steps_per_second": 6.17,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 46.64506530761719,
+      "learning_rate": 4.777896233120114e-06,
+      "loss": 1.1393,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 41.11891174316406,
+      "learning_rate": 4.733475479744136e-06,
+      "loss": 1.0243,
+      "step": 150
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 34.20009994506836,
+      "learning_rate": 4.6890547263681595e-06,
+      "loss": 0.9005,
+      "step": 175
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 38.80377197265625,
+      "learning_rate": 4.644633972992183e-06,
+      "loss": 0.8627,
+      "step": 200
+    },
+    {
+      "epoch": 0.21,
+      "eval_accuracy": 0.7907,
+      "eval_f1_macro": 0.6657039157603262,
+      "eval_f1_micro": 0.7907,
+      "eval_loss": 0.8368468880653381,
+      "eval_runtime": 25.9484,
+      "eval_samples_per_second": 385.38,
+      "eval_steps_per_second": 6.05,
+      "step": 200
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 34.973506927490234,
+      "learning_rate": 4.600213219616206e-06,
+      "loss": 0.7896,
+      "step": 225
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 29.98388671875,
+      "learning_rate": 4.555792466240228e-06,
+      "loss": 0.8307,
+      "step": 250
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 27.09973907470703,
+      "learning_rate": 4.51137171286425e-06,
+      "loss": 0.7846,
+      "step": 275
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 36.151161193847656,
+      "learning_rate": 4.466950959488273e-06,
+      "loss": 0.7179,
+      "step": 300
+    },
+    {
+      "epoch": 0.32,
+      "eval_accuracy": 0.7971,
+      "eval_f1_macro": 0.6861778340669753,
+      "eval_f1_micro": 0.7971,
+      "eval_loss": 0.7824062705039978,
+      "eval_runtime": 25.9003,
+      "eval_samples_per_second": 386.095,
+      "eval_steps_per_second": 6.062,
+      "step": 300
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 33.09822463989258,
+      "learning_rate": 4.422530206112296e-06,
+      "loss": 0.7133,
+      "step": 325
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 35.52923583984375,
+      "learning_rate": 4.378109452736319e-06,
+      "loss": 0.7354,
+      "step": 350
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 39.79545211791992,
+      "learning_rate": 4.333688699360342e-06,
+      "loss": 0.6619,
+      "step": 375
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 30.13161849975586,
+      "learning_rate": 4.289267945984365e-06,
+      "loss": 0.6961,
+      "step": 400
+    },
+    {
+      "epoch": 0.43,
+      "eval_accuracy": 0.8138,
+      "eval_f1_macro": 0.6992465625213966,
+      "eval_f1_micro": 0.8138,
+      "eval_loss": 0.6951531171798706,
+      "eval_runtime": 25.6082,
+      "eval_samples_per_second": 390.5,
+      "eval_steps_per_second": 6.131,
+      "step": 400
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 27.575519561767578,
+      "learning_rate": 4.244847192608387e-06,
+      "loss": 0.7162,
+      "step": 425
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 35.084754943847656,
+      "learning_rate": 4.200426439232409e-06,
+      "loss": 0.7722,
+      "step": 450
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 28.47511863708496,
+      "learning_rate": 4.156005685856432e-06,
+      "loss": 0.6866,
+      "step": 475
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 32.34709548950195,
+      "learning_rate": 4.1115849324804554e-06,
+      "loss": 0.745,
+      "step": 500
+    },
+    {
+      "epoch": 0.53,
+      "eval_accuracy": 0.8121,
+      "eval_f1_macro": 0.7033560293953169,
+      "eval_f1_micro": 0.8121,
+      "eval_loss": 0.6718780994415283,
+      "eval_runtime": 25.9161,
+      "eval_samples_per_second": 385.86,
+      "eval_steps_per_second": 6.058,
+      "step": 500
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 25.5845890045166,
+      "learning_rate": 4.067164179104478e-06,
+      "loss": 0.6535,
+      "step": 525
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 22.466503143310547,
+      "learning_rate": 4.022743425728501e-06,
+      "loss": 0.5969,
+      "step": 550
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 27.53134536743164,
+      "learning_rate": 3.978322672352524e-06,
+      "loss": 0.5926,
+      "step": 575
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 31.356454849243164,
+      "learning_rate": 3.933901918976546e-06,
+      "loss": 0.6505,
+      "step": 600
+    },
+    {
+      "epoch": 0.64,
+      "eval_accuracy": 0.834,
+      "eval_f1_macro": 0.7469091035082649,
+      "eval_f1_micro": 0.834,
+      "eval_loss": 0.6219750046730042,
+      "eval_runtime": 25.9316,
+      "eval_samples_per_second": 385.63,
+      "eval_steps_per_second": 6.054,
+      "step": 600
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 37.17654800415039,
+      "learning_rate": 3.889481165600569e-06,
+      "loss": 0.6171,
+      "step": 625
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 26.71038055419922,
+      "learning_rate": 3.8450604122245914e-06,
+      "loss": 0.6218,
+      "step": 650
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 27.787952423095703,
+      "learning_rate": 3.8006396588486145e-06,
+      "loss": 0.6124,
+      "step": 675
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 30.405912399291992,
+      "learning_rate": 3.756218905472637e-06,
+      "loss": 0.5914,
+      "step": 700
+    },
+    {
+      "epoch": 0.75,
+      "eval_accuracy": 0.8362,
+      "eval_f1_macro": 0.7410957777496914,
+      "eval_f1_micro": 0.8362,
+      "eval_loss": 0.6109625101089478,
+      "eval_runtime": 25.6247,
+      "eval_samples_per_second": 390.248,
+      "eval_steps_per_second": 6.127,
+      "step": 700
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 30.52012062072754,
+      "learning_rate": 3.71179815209666e-06,
+      "loss": 0.5711,
+      "step": 725
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 30.88004493713379,
+      "learning_rate": 3.667377398720683e-06,
+      "loss": 0.6695,
+      "step": 750
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 22.504459381103516,
+      "learning_rate": 3.622956645344705e-06,
+      "loss": 0.5731,
+      "step": 775
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 21.515512466430664,
+      "learning_rate": 3.578535891968728e-06,
+      "loss": 0.5837,
+      "step": 800
+    },
+    {
+      "epoch": 0.85,
+      "eval_accuracy": 0.8385,
+      "eval_f1_macro": 0.7413235492734335,
+      "eval_f1_micro": 0.8385,
+      "eval_loss": 0.5766780972480774,
+      "eval_runtime": 25.6608,
+      "eval_samples_per_second": 389.7,
+      "eval_steps_per_second": 6.118,
+      "step": 800
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 30.9660587310791,
+      "learning_rate": 3.534115138592751e-06,
+      "loss": 0.6085,
+      "step": 825
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 18.883647918701172,
+      "learning_rate": 3.4896943852167736e-06,
+      "loss": 0.5121,
+      "step": 850
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 24.548561096191406,
+      "learning_rate": 3.4452736318407963e-06,
+      "loss": 0.5621,
+      "step": 875
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 29.833791732788086,
+      "learning_rate": 3.4008528784648194e-06,
+      "loss": 0.5218,
+      "step": 900
+    },
+    {
+      "epoch": 0.96,
+      "eval_accuracy": 0.849,
+      "eval_f1_macro": 0.7702797685808792,
+      "eval_f1_micro": 0.849,
+      "eval_loss": 0.5365203022956848,
+      "eval_runtime": 25.9091,
+      "eval_samples_per_second": 385.964,
+      "eval_steps_per_second": 6.06,
+      "step": 900
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 27.948928833007812,
+      "learning_rate": 3.3564321250888416e-06,
+      "loss": 0.5681,
+      "step": 925
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 19.800880432128906,
+      "learning_rate": 3.3120113717128643e-06,
+      "loss": 0.4014,
+      "step": 950
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 19.333465576171875,
+      "learning_rate": 3.2675906183368874e-06,
+      "loss": 0.2795,
+      "step": 975
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 22.315195083618164,
+      "learning_rate": 3.22316986496091e-06,
+      "loss": 0.2632,
+      "step": 1000
+    },
+    {
+      "epoch": 1.07,
+      "eval_accuracy": 0.8562,
+      "eval_f1_macro": 0.7683569808757446,
+      "eval_f1_micro": 0.8562,
+      "eval_loss": 0.5503664016723633,
+      "eval_runtime": 25.5198,
+      "eval_samples_per_second": 391.852,
+      "eval_steps_per_second": 6.152,
+      "step": 1000
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 24.819501876831055,
+      "learning_rate": 3.1787491115849327e-06,
+      "loss": 0.2532,
+      "step": 1025
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 21.534936904907227,
+      "learning_rate": 3.1343283582089558e-06,
+      "loss": 0.2311,
+      "step": 1050
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 24.088809967041016,
+      "learning_rate": 3.0899076048329785e-06,
+      "loss": 0.3134,
+      "step": 1075
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 27.605493545532227,
+      "learning_rate": 3.0454868514570007e-06,
+      "loss": 0.2607,
+      "step": 1100
+    },
+    {
+      "epoch": 1.17,
+      "eval_accuracy": 0.8525,
+      "eval_f1_macro": 0.7656891626030512,
+      "eval_f1_micro": 0.8525,
+      "eval_loss": 0.5496523380279541,
+      "eval_runtime": 25.7081,
+      "eval_samples_per_second": 388.982,
+      "eval_steps_per_second": 6.107,
+      "step": 1100
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 22.955158233642578,
+      "learning_rate": 3.0010660980810234e-06,
+      "loss": 0.2674,
+      "step": 1125
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 19.089893341064453,
+      "learning_rate": 2.9566453447050464e-06,
+      "loss": 0.2074,
+      "step": 1150
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 19.285688400268555,
+      "learning_rate": 2.912224591329069e-06,
+      "loss": 0.2488,
+      "step": 1175
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 23.45233726501465,
+      "learning_rate": 2.867803837953092e-06,
+      "loss": 0.274,
+      "step": 1200
+    },
+    {
+      "epoch": 1.28,
+      "eval_accuracy": 0.8584,
+      "eval_f1_macro": 0.7746299057445165,
+      "eval_f1_micro": 0.8584,
+      "eval_loss": 0.5439000129699707,
+      "eval_runtime": 25.9014,
+      "eval_samples_per_second": 386.079,
+      "eval_steps_per_second": 6.061,
+      "step": 1200
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 31.231454849243164,
+      "learning_rate": 2.823383084577115e-06,
+      "loss": 0.2624,
+      "step": 1225
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 28.1010799407959,
+      "learning_rate": 2.7789623312011375e-06,
+      "loss": 0.2992,
+      "step": 1250
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 30.002384185791016,
+      "learning_rate": 2.7345415778251598e-06,
+      "loss": 0.2589,
+      "step": 1275
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 23.61323356628418,
+      "learning_rate": 2.690120824449183e-06,
+      "loss": 0.2216,
+      "step": 1300
+    },
+    {
+      "epoch": 1.39,
+      "eval_accuracy": 0.8563,
+      "eval_f1_macro": 0.7753520513346309,
+      "eval_f1_micro": 0.8563,
+      "eval_loss": 0.5687375068664551,
+      "eval_runtime": 25.9424,
+      "eval_samples_per_second": 385.47,
+      "eval_steps_per_second": 6.052,
+      "step": 1300
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 27.56183433532715,
+      "learning_rate": 2.6457000710732055e-06,
+      "loss": 0.2845,
+      "step": 1325
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 18.88576316833496,
+      "learning_rate": 2.601279317697228e-06,
+      "loss": 0.2685,
+      "step": 1350
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 19.662220001220703,
+      "learning_rate": 2.5568585643212513e-06,
+      "loss": 0.2489,
+      "step": 1375
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 22.736656188964844,
+      "learning_rate": 2.512437810945274e-06,
+      "loss": 0.2044,
+      "step": 1400
+    },
+    {
+      "epoch": 1.49,
+      "eval_accuracy": 0.861,
+      "eval_f1_macro": 0.7820141563614671,
+      "eval_f1_micro": 0.861,
+      "eval_loss": 0.5385035276412964,
+      "eval_runtime": 25.6666,
+      "eval_samples_per_second": 389.612,
+      "eval_steps_per_second": 6.117,
+      "step": 1400
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 24.569435119628906,
+      "learning_rate": 2.4680170575692966e-06,
+      "loss": 0.2388,
+      "step": 1425
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 17.50179100036621,
+      "learning_rate": 2.4235963041933193e-06,
+      "loss": 0.2556,
+      "step": 1450
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 15.387917518615723,
+      "learning_rate": 2.379175550817342e-06,
+      "loss": 0.2343,
+      "step": 1475
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 29.757495880126953,
+      "learning_rate": 2.3347547974413646e-06,
+      "loss": 0.2508,
+      "step": 1500
+    },
+    {
+      "epoch": 1.6,
+      "eval_accuracy": 0.8577,
+      "eval_f1_macro": 0.7710712973870113,
+      "eval_f1_micro": 0.8577,
+      "eval_loss": 0.5657808780670166,
+      "eval_runtime": 25.9754,
+      "eval_samples_per_second": 384.98,
+      "eval_steps_per_second": 6.044,
+      "step": 1500
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 24.104217529296875,
+      "learning_rate": 2.2903340440653877e-06,
+      "loss": 0.2647,
+      "step": 1525
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 29.48048973083496,
+      "learning_rate": 2.24591329068941e-06,
+      "loss": 0.212,
+      "step": 1550
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 11.834880828857422,
+      "learning_rate": 2.201492537313433e-06,
+      "loss": 0.1939,
+      "step": 1575
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 24.24506378173828,
+      "learning_rate": 2.1570717839374557e-06,
+      "loss": 0.2513,
+      "step": 1600
+    },
+    {
+      "epoch": 1.71,
+      "eval_accuracy": 0.8589,
+      "eval_f1_macro": 0.7871987440671023,
+      "eval_f1_micro": 0.8589,
+      "eval_loss": 0.5366827845573425,
+      "eval_runtime": 25.9643,
+      "eval_samples_per_second": 385.144,
+      "eval_steps_per_second": 6.047,
+      "step": 1600
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 23.33180046081543,
+      "learning_rate": 2.112651030561479e-06,
+      "loss": 0.2409,
+      "step": 1625
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 18.71114730834961,
+      "learning_rate": 2.068230277185501e-06,
+      "loss": 0.224,
+      "step": 1650
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 21.95819854736328,
+      "learning_rate": 2.023809523809524e-06,
+      "loss": 0.2223,
+      "step": 1675
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 27.065677642822266,
+      "learning_rate": 1.979388770433547e-06,
+      "loss": 0.2787,
+      "step": 1700
+    },
+    {
+      "epoch": 1.81,
+      "eval_accuracy": 0.8653,
+      "eval_f1_macro": 0.790261134849528,
+      "eval_f1_micro": 0.8653,
+      "eval_loss": 0.5133171677589417,
+      "eval_runtime": 25.5701,
+      "eval_samples_per_second": 391.081,
+      "eval_steps_per_second": 6.14,
+      "step": 1700
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 35.288761138916016,
+      "learning_rate": 1.9349680170575695e-06,
+      "loss": 0.2709,
+      "step": 1725
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 21.077306747436523,
+      "learning_rate": 1.8905472636815921e-06,
+      "loss": 0.2002,
+      "step": 1750
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 25.394838333129883,
+      "learning_rate": 1.846126510305615e-06,
+      "loss": 0.2461,
+      "step": 1775
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 26.597759246826172,
+      "learning_rate": 1.8017057569296375e-06,
+      "loss": 0.2357,
+      "step": 1800
+    },
+    {
+      "epoch": 1.92,
+      "eval_accuracy": 0.8669,
+      "eval_f1_macro": 0.7902403947168268,
+      "eval_f1_micro": 0.8669,
+      "eval_loss": 0.5063937306404114,
+      "eval_runtime": 25.6031,
+      "eval_samples_per_second": 390.577,
+      "eval_steps_per_second": 6.132,
+      "step": 1800
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 2814,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "total_flos": 1.074692042564567e+17,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}

Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f26a8af75d31668209ed94c987a4170cdf12fb1b08369f54b4ac17f6f18edc56
+size 5944

Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800/zero_to_fp32.py

@@ -0,0 +1,604 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example: python zero_to_fp32.py . pytorch_model.bin
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage <= 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+
+    total_files = len(files)
+    state_dicts = []
+    for f in files:
+        state_dict = torch.load(f, map_location=device)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    if zero_stage <= 2:
+        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    elif zero_stage == 3:
+        # if there is more than one param group, there will be multiple flattened tensors - one
+        # flattened tensor per group - for simplicity merge them into a single tensor
+        #
+        # XXX: could make the script more memory efficient for when there are multiple groups - it
+        # will require matching the sub-lists of param_shapes for each param group flattened tensor
+
+        fp32_flat_groups = [
+            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
+        ]
+
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = fp32_flat_groups[0].numel() * world_size
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    for name, shape in param_shapes.items():
+
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # XXX: memory usage doubles here
+        state_dict[name] = torch.cat(
+            tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
+            0).narrow(0, 0, unpartitioned_numel).view(shape)
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+
+    Returns:
+        - pytorch ``state_dict``
+
+    Note: this approach may not work if your application doesn't have sufficient free CPU memory and
+    you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+    """
+
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
+    print(f"Saving fp32 state dict to {output_file}")
+    torch.save(state_dict, output_file)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info(f"Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info(f"Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument(
+        "output_file",
+        type=str,
+        help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
+                                               args.output_file,
+                                               tag=args.tag,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)

Qwen/Qwen1.5_1.8B_ledgar/config.json

@@ -0,0 +1,235 @@
+{
+  "_name_or_path": "Qwen/Qwen1.5-1.8B",
+  "architectures": [
+    "Qwen2ForSequenceClassification"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "finetuning_task": "text-classification",
+  "hidden_act": "silu",
+  "hidden_size": 2048,
+  "id2label": {
+    "0": "0",
+    "1": "1",
+    "2": "10",
+    "3": "11",
+    "4": "12",
+    "5": "13",
+    "6": "14",
+    "7": "15",
+    "8": "16",
+    "9": "17",
+    "10": "18",
+    "11": "19",
+    "12": "2",
+    "13": "20",
+    "14": "21",
+    "15": "22",
+    "16": "23",
+    "17": "24",
+    "18": "25",
+    "19": "26",
+    "20": "27",
+    "21": "28",
+    "22": "29",
+    "23": "3",
+    "24": "30",
+    "25": "31",
+    "26": "32",
+    "27": "33",
+    "28": "34",
+    "29": "35",
+    "30": "36",
+    "31": "37",
+    "32": "38",
+    "33": "39",
+    "34": "4",
+    "35": "40",
+    "36": "41",
+    "37": "42",
+    "38": "43",
+    "39": "44",
+    "40": "45",
+    "41": "46",
+    "42": "47",
+    "43": "48",
+    "44": "49",
+    "45": "5",
+    "46": "50",
+    "47": "51",
+    "48": "52",
+    "49": "53",
+    "50": "54",
+    "51": "55",
+    "52": "56",
+    "53": "57",
+    "54": "58",
+    "55": "59",
+    "56": "6",
+    "57": "60",
+    "58": "61",
+    "59": "62",
+    "60": "63",
+    "61": "64",
+    "62": "65",
+    "63": "66",
+    "64": "67",
+    "65": "68",
+    "66": "69",
+    "67": "7",
+    "68": "70",
+    "69": "71",
+    "70": "72",
+    "71": "73",
+    "72": "74",
+    "73": "75",
+    "74": "76",
+    "75": "77",
+    "76": "78",
+    "77": "79",
+    "78": "8",
+    "79": "80",
+    "80": "81",
+    "81": "82",
+    "82": "83",
+    "83": "84",
+    "84": "85",
+    "85": "86",
+    "86": "87",
+    "87": "88",
+    "88": "89",
+    "89": "9",
+    "90": "90",
+    "91": "91",
+    "92": "92",
+    "93": "93",
+    "94": "94",
+    "95": "95",
+    "96": "96",
+    "97": "97",
+    "98": "98",
+    "99": "99"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 5504,
+  "label2id": {
+    "0": 0,
+    "1": 1,
+    "10": 2,
+    "11": 3,
+    "12": 4,
+    "13": 5,
+    "14": 6,
+    "15": 7,
+    "16": 8,
+    "17": 9,
+    "18": 10,
+    "19": 11,
+    "2": 12,
+    "20": 13,
+    "21": 14,
+    "22": 15,
+    "23": 16,
+    "24": 17,
+    "25": 18,
+    "26": 19,
+    "27": 20,
+    "28": 21,
+    "29": 22,
+    "3": 23,
+    "30": 24,
+    "31": 25,
+    "32": 26,
+    "33": 27,
+    "34": 28,
+    "35": 29,
+    "36": 30,
+    "37": 31,
+    "38": 32,
+    "39": 33,
+    "4": 34,
+    "40": 35,
+    "41": 36,
+    "42": 37,
+    "43": 38,
+    "44": 39,
+    "45": 40,
+    "46": 41,
+    "47": 42,
+    "48": 43,
+    "49": 44,
+    "5": 45,
+    "50": 46,
+    "51": 47,
+    "52": 48,
+    "53": 49,
+    "54": 50,
+    "55": 51,
+    "56": 52,
+    "57": 53,
+    "58": 54,
+    "59": 55,
+    "6": 56,
+    "60": 57,
+    "61": 58,
+    "62": 59,
+    "63": 60,
+    "64": 61,
+    "65": 62,
+    "66": 63,
+    "67": 64,
+    "68": 65,
+    "69": 66,
+    "7": 67,
+    "70": 68,
+    "71": 69,
+    "72": 70,
+    "73": 71,
+    "74": 72,
+    "75": 73,
+    "76": 74,
+    "77": 75,
+    "78": 76,
+    "79": 77,
+    "8": 78,
+    "80": 79,
+    "81": 80,
+    "82": 81,
+    "83": 82,
+    "84": 83,
+    "85": 84,
+    "86": 85,
+    "87": 86,
+    "88": 87,
+    "89": 88,
+    "9": 89,
+    "90": 90,
+    "91": 91,
+    "92": 92,
+    "93": 93,
+    "94": 94,
+    "95": 95,
+    "96": 96,
+    "97": 97,
+    "98": 98,
+    "99": 99
+  },
+  "max_position_embeddings": 32768,
+  "max_window_layers": 21,
+  "model_type": "qwen2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 24,
+  "num_key_value_heads": 16,
+  "pad_token_id": 151643,
+  "problem_type": "single_label_classification",
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 1000000.0,
+  "sliding_window": 32768,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.39.0.dev0",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151646
+}

Qwen/Qwen1.5_1.8B_ledgar/eval_results.json

@@ -0,0 +1,11 @@
+{
+    "epoch": 3.0,
+    "eval_accuracy": 0.8669,
+    "eval_f1_macro": 0.7902403947168268,
+    "eval_f1_micro": 0.8669,
+    "eval_loss": 0.5063937306404114,
+    "eval_runtime": 24.4305,
+    "eval_samples": 10000,
+    "eval_samples_per_second": 409.324,
+    "eval_steps_per_second": 6.426
+}

Qwen/Qwen1.5_1.8B_ledgar/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6be1da9a3d742bcd4f56af0dd627ec82b2b529a0f3a3259561a3da60f4f2dffc
+size 3050582504

Qwen/Qwen1.5_1.8B_ledgar/run.log

@@ -0,0 +1,4 @@
+03/16/2024 00:29:25 - WARNING - __main__ - Process rank: 0, device: cuda:0, n_gpu: 1, distributed training: True, 16-bits training: False
+03/16/2024 00:29:26 - WARNING - __main__ - Process rank: 1, device: cuda:1, n_gpu: 1, distributed training: True, 16-bits training: False
+03/16/2024 00:29:35 - WARNING - __main__ - The label2id key in the model config.json is not equal to the label2id key of this run. You can ignore this if you are doing finetuning.
+03/16/2024 00:29:36 - WARNING - __main__ - The label2id key in the model config.json is not equal to the label2id key of this run. You can ignore this if you are doing finetuning.

Qwen/Qwen1.5_1.8B_ledgar/special_tokens_map.json

@@ -0,0 +1,14 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>"
+  ],
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<|endoftext|>"
+}

Qwen/Qwen1.5_1.8B_ledgar/test_results.json

@@ -0,0 +1,10 @@
+{
+    "epoch": 3.0,
+    "test_accuracy": 0.8664,
+    "test_f1_macro": 0.7974226514742132,
+    "test_f1_micro": 0.8664,
+    "test_loss": 0.532435953617096,
+    "test_runtime": 25.4113,
+    "test_samples_per_second": 393.525,
+    "test_steps_per_second": 6.178
+}

Qwen/Qwen1.5_1.8B_ledgar/tokenizer_config.json

@@ -0,0 +1,43 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{% for message in messages %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "model_max_length": 32768,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

Qwen/Qwen1.5_1.8B_ledgar/train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 3.0,
+    "train_loss": 0.42635450247932005,
+    "train_runtime": 3623.2348,
+    "train_samples": 60000,
+    "train_samples_per_second": 49.679,
+    "train_steps_per_second": 0.777
+}

Qwen/Qwen1.5_1.8B_ledgar/trainer_state.json

@@ -0,0 +1,1122 @@
+{
+  "best_metric": 0.5063937306404114,
+  "best_model_checkpoint": "../../experiments_checkpoints/MAdAiLab/Qwen/Qwen1.5_1.8B_ledgar/checkpoint-1800",
+  "epoch": 3.0,
+  "eval_steps": 100,
+  "global_step": 2814,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.03,
+      "grad_norm": 80.4836196899414,
+      "learning_rate": 4.9555792466240235e-06,
+      "loss": 7.9887,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 60.927364349365234,
+      "learning_rate": 4.911158493248046e-06,
+      "loss": 3.1908,
+      "step": 50
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 49.68091583251953,
+      "learning_rate": 4.866737739872069e-06,
+      "loss": 1.7183,
+      "step": 75
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 55.176666259765625,
+      "learning_rate": 4.822316986496091e-06,
+      "loss": 1.3077,
+      "step": 100
+    },
+    {
+      "epoch": 0.11,
+      "eval_accuracy": 0.7277,
+      "eval_f1_macro": 0.5770831474844406,
+      "eval_f1_micro": 0.7277,
+      "eval_loss": 1.0944937467575073,
+      "eval_runtime": 25.447,
+      "eval_samples_per_second": 392.973,
+      "eval_steps_per_second": 6.17,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 46.64506530761719,
+      "learning_rate": 4.777896233120114e-06,
+      "loss": 1.1393,
+      "step": 125
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 41.11891174316406,
+      "learning_rate": 4.733475479744136e-06,
+      "loss": 1.0243,
+      "step": 150
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 34.20009994506836,
+      "learning_rate": 4.6890547263681595e-06,
+      "loss": 0.9005,
+      "step": 175
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 38.80377197265625,
+      "learning_rate": 4.644633972992183e-06,
+      "loss": 0.8627,
+      "step": 200
+    },
+    {
+      "epoch": 0.21,
+      "eval_accuracy": 0.7907,
+      "eval_f1_macro": 0.6657039157603262,
+      "eval_f1_micro": 0.7907,
+      "eval_loss": 0.8368468880653381,
+      "eval_runtime": 25.9484,
+      "eval_samples_per_second": 385.38,
+      "eval_steps_per_second": 6.05,
+      "step": 200
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 34.973506927490234,
+      "learning_rate": 4.600213219616206e-06,
+      "loss": 0.7896,
+      "step": 225
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 29.98388671875,
+      "learning_rate": 4.555792466240228e-06,
+      "loss": 0.8307,
+      "step": 250
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 27.09973907470703,
+      "learning_rate": 4.51137171286425e-06,
+      "loss": 0.7846,
+      "step": 275
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 36.151161193847656,
+      "learning_rate": 4.466950959488273e-06,
+      "loss": 0.7179,
+      "step": 300
+    },
+    {
+      "epoch": 0.32,
+      "eval_accuracy": 0.7971,
+      "eval_f1_macro": 0.6861778340669753,
+      "eval_f1_micro": 0.7971,
+      "eval_loss": 0.7824062705039978,
+      "eval_runtime": 25.9003,
+      "eval_samples_per_second": 386.095,
+      "eval_steps_per_second": 6.062,
+      "step": 300
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 33.09822463989258,
+      "learning_rate": 4.422530206112296e-06,
+      "loss": 0.7133,
+      "step": 325
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 35.52923583984375,
+      "learning_rate": 4.378109452736319e-06,
+      "loss": 0.7354,
+      "step": 350
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 39.79545211791992,
+      "learning_rate": 4.333688699360342e-06,
+      "loss": 0.6619,
+      "step": 375
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 30.13161849975586,
+      "learning_rate": 4.289267945984365e-06,
+      "loss": 0.6961,
+      "step": 400
+    },
+    {
+      "epoch": 0.43,
+      "eval_accuracy": 0.8138,
+      "eval_f1_macro": 0.6992465625213966,
+      "eval_f1_micro": 0.8138,
+      "eval_loss": 0.6951531171798706,
+      "eval_runtime": 25.6082,
+      "eval_samples_per_second": 390.5,
+      "eval_steps_per_second": 6.131,
+      "step": 400
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 27.575519561767578,
+      "learning_rate": 4.244847192608387e-06,
+      "loss": 0.7162,
+      "step": 425
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 35.084754943847656,
+      "learning_rate": 4.200426439232409e-06,
+      "loss": 0.7722,
+      "step": 450
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 28.47511863708496,
+      "learning_rate": 4.156005685856432e-06,
+      "loss": 0.6866,
+      "step": 475
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 32.34709548950195,
+      "learning_rate": 4.1115849324804554e-06,
+      "loss": 0.745,
+      "step": 500
+    },
+    {
+      "epoch": 0.53,
+      "eval_accuracy": 0.8121,
+      "eval_f1_macro": 0.7033560293953169,
+      "eval_f1_micro": 0.8121,
+      "eval_loss": 0.6718780994415283,
+      "eval_runtime": 25.9161,
+      "eval_samples_per_second": 385.86,
+      "eval_steps_per_second": 6.058,
+      "step": 500
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 25.5845890045166,
+      "learning_rate": 4.067164179104478e-06,
+      "loss": 0.6535,
+      "step": 525
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 22.466503143310547,
+      "learning_rate": 4.022743425728501e-06,
+      "loss": 0.5969,
+      "step": 550
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 27.53134536743164,
+      "learning_rate": 3.978322672352524e-06,
+      "loss": 0.5926,
+      "step": 575
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 31.356454849243164,
+      "learning_rate": 3.933901918976546e-06,
+      "loss": 0.6505,
+      "step": 600
+    },
+    {
+      "epoch": 0.64,
+      "eval_accuracy": 0.834,
+      "eval_f1_macro": 0.7469091035082649,
+      "eval_f1_micro": 0.834,
+      "eval_loss": 0.6219750046730042,
+      "eval_runtime": 25.9316,
+      "eval_samples_per_second": 385.63,
+      "eval_steps_per_second": 6.054,
+      "step": 600
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 37.17654800415039,
+      "learning_rate": 3.889481165600569e-06,
+      "loss": 0.6171,
+      "step": 625
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 26.71038055419922,
+      "learning_rate": 3.8450604122245914e-06,
+      "loss": 0.6218,
+      "step": 650
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 27.787952423095703,
+      "learning_rate": 3.8006396588486145e-06,
+      "loss": 0.6124,
+      "step": 675
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 30.405912399291992,
+      "learning_rate": 3.756218905472637e-06,
+      "loss": 0.5914,
+      "step": 700
+    },
+    {
+      "epoch": 0.75,
+      "eval_accuracy": 0.8362,
+      "eval_f1_macro": 0.7410957777496914,
+      "eval_f1_micro": 0.8362,
+      "eval_loss": 0.6109625101089478,
+      "eval_runtime": 25.6247,
+      "eval_samples_per_second": 390.248,
+      "eval_steps_per_second": 6.127,
+      "step": 700
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 30.52012062072754,
+      "learning_rate": 3.71179815209666e-06,
+      "loss": 0.5711,
+      "step": 725
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 30.88004493713379,
+      "learning_rate": 3.667377398720683e-06,
+      "loss": 0.6695,
+      "step": 750
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 22.504459381103516,
+      "learning_rate": 3.622956645344705e-06,
+      "loss": 0.5731,
+      "step": 775
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 21.515512466430664,
+      "learning_rate": 3.578535891968728e-06,
+      "loss": 0.5837,
+      "step": 800
+    },
+    {
+      "epoch": 0.85,
+      "eval_accuracy": 0.8385,
+      "eval_f1_macro": 0.7413235492734335,
+      "eval_f1_micro": 0.8385,
+      "eval_loss": 0.5766780972480774,
+      "eval_runtime": 25.6608,
+      "eval_samples_per_second": 389.7,
+      "eval_steps_per_second": 6.118,
+      "step": 800
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 30.9660587310791,
+      "learning_rate": 3.534115138592751e-06,
+      "loss": 0.6085,
+      "step": 825
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 18.883647918701172,
+      "learning_rate": 3.4896943852167736e-06,
+      "loss": 0.5121,
+      "step": 850
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 24.548561096191406,
+      "learning_rate": 3.4452736318407963e-06,
+      "loss": 0.5621,
+      "step": 875
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 29.833791732788086,
+      "learning_rate": 3.4008528784648194e-06,
+      "loss": 0.5218,
+      "step": 900
+    },
+    {
+      "epoch": 0.96,
+      "eval_accuracy": 0.849,
+      "eval_f1_macro": 0.7702797685808792,
+      "eval_f1_micro": 0.849,
+      "eval_loss": 0.5365203022956848,
+      "eval_runtime": 25.9091,
+      "eval_samples_per_second": 385.964,
+      "eval_steps_per_second": 6.06,
+      "step": 900
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 27.948928833007812,
+      "learning_rate": 3.3564321250888416e-06,
+      "loss": 0.5681,
+      "step": 925
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 19.800880432128906,
+      "learning_rate": 3.3120113717128643e-06,
+      "loss": 0.4014,
+      "step": 950
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 19.333465576171875,
+      "learning_rate": 3.2675906183368874e-06,
+      "loss": 0.2795,
+      "step": 975
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 22.315195083618164,
+      "learning_rate": 3.22316986496091e-06,
+      "loss": 0.2632,
+      "step": 1000
+    },
+    {
+      "epoch": 1.07,
+      "eval_accuracy": 0.8562,
+      "eval_f1_macro": 0.7683569808757446,
+      "eval_f1_micro": 0.8562,
+      "eval_loss": 0.5503664016723633,
+      "eval_runtime": 25.5198,
+      "eval_samples_per_second": 391.852,
+      "eval_steps_per_second": 6.152,
+      "step": 1000
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 24.819501876831055,
+      "learning_rate": 3.1787491115849327e-06,
+      "loss": 0.2532,
+      "step": 1025
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 21.534936904907227,
+      "learning_rate": 3.1343283582089558e-06,
+      "loss": 0.2311,
+      "step": 1050
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 24.088809967041016,
+      "learning_rate": 3.0899076048329785e-06,
+      "loss": 0.3134,
+      "step": 1075
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 27.605493545532227,
+      "learning_rate": 3.0454868514570007e-06,
+      "loss": 0.2607,
+      "step": 1100
+    },
+    {
+      "epoch": 1.17,
+      "eval_accuracy": 0.8525,
+      "eval_f1_macro": 0.7656891626030512,
+      "eval_f1_micro": 0.8525,
+      "eval_loss": 0.5496523380279541,
+      "eval_runtime": 25.7081,
+      "eval_samples_per_second": 388.982,
+      "eval_steps_per_second": 6.107,
+      "step": 1100
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 22.955158233642578,
+      "learning_rate": 3.0010660980810234e-06,
+      "loss": 0.2674,
+      "step": 1125
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 19.089893341064453,
+      "learning_rate": 2.9566453447050464e-06,
+      "loss": 0.2074,
+      "step": 1150
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 19.285688400268555,
+      "learning_rate": 2.912224591329069e-06,
+      "loss": 0.2488,
+      "step": 1175
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 23.45233726501465,
+      "learning_rate": 2.867803837953092e-06,
+      "loss": 0.274,
+      "step": 1200
+    },
+    {
+      "epoch": 1.28,
+      "eval_accuracy": 0.8584,
+      "eval_f1_macro": 0.7746299057445165,
+      "eval_f1_micro": 0.8584,
+      "eval_loss": 0.5439000129699707,
+      "eval_runtime": 25.9014,
+      "eval_samples_per_second": 386.079,
+      "eval_steps_per_second": 6.061,
+      "step": 1200
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 31.231454849243164,
+      "learning_rate": 2.823383084577115e-06,
+      "loss": 0.2624,
+      "step": 1225
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 28.1010799407959,
+      "learning_rate": 2.7789623312011375e-06,
+      "loss": 0.2992,
+      "step": 1250
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 30.002384185791016,
+      "learning_rate": 2.7345415778251598e-06,
+      "loss": 0.2589,
+      "step": 1275
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 23.61323356628418,
+      "learning_rate": 2.690120824449183e-06,
+      "loss": 0.2216,
+      "step": 1300
+    },
+    {
+      "epoch": 1.39,
+      "eval_accuracy": 0.8563,
+      "eval_f1_macro": 0.7753520513346309,
+      "eval_f1_micro": 0.8563,
+      "eval_loss": 0.5687375068664551,
+      "eval_runtime": 25.9424,
+      "eval_samples_per_second": 385.47,
+      "eval_steps_per_second": 6.052,
+      "step": 1300
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 27.56183433532715,
+      "learning_rate": 2.6457000710732055e-06,
+      "loss": 0.2845,
+      "step": 1325
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 18.88576316833496,
+      "learning_rate": 2.601279317697228e-06,
+      "loss": 0.2685,
+      "step": 1350
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 19.662220001220703,
+      "learning_rate": 2.5568585643212513e-06,
+      "loss": 0.2489,
+      "step": 1375
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 22.736656188964844,
+      "learning_rate": 2.512437810945274e-06,
+      "loss": 0.2044,
+      "step": 1400
+    },
+    {
+      "epoch": 1.49,
+      "eval_accuracy": 0.861,
+      "eval_f1_macro": 0.7820141563614671,
+      "eval_f1_micro": 0.861,
+      "eval_loss": 0.5385035276412964,
+      "eval_runtime": 25.6666,
+      "eval_samples_per_second": 389.612,
+      "eval_steps_per_second": 6.117,
+      "step": 1400
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 24.569435119628906,
+      "learning_rate": 2.4680170575692966e-06,
+      "loss": 0.2388,
+      "step": 1425
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 17.50179100036621,
+      "learning_rate": 2.4235963041933193e-06,
+      "loss": 0.2556,
+      "step": 1450
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 15.387917518615723,
+      "learning_rate": 2.379175550817342e-06,
+      "loss": 0.2343,
+      "step": 1475
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 29.757495880126953,
+      "learning_rate": 2.3347547974413646e-06,
+      "loss": 0.2508,
+      "step": 1500
+    },
+    {
+      "epoch": 1.6,
+      "eval_accuracy": 0.8577,
+      "eval_f1_macro": 0.7710712973870113,
+      "eval_f1_micro": 0.8577,
+      "eval_loss": 0.5657808780670166,
+      "eval_runtime": 25.9754,
+      "eval_samples_per_second": 384.98,
+      "eval_steps_per_second": 6.044,
+      "step": 1500
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 24.104217529296875,
+      "learning_rate": 2.2903340440653877e-06,
+      "loss": 0.2647,
+      "step": 1525
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 29.48048973083496,
+      "learning_rate": 2.24591329068941e-06,
+      "loss": 0.212,
+      "step": 1550
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 11.834880828857422,
+      "learning_rate": 2.201492537313433e-06,
+      "loss": 0.1939,
+      "step": 1575
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 24.24506378173828,
+      "learning_rate": 2.1570717839374557e-06,
+      "loss": 0.2513,
+      "step": 1600
+    },
+    {
+      "epoch": 1.71,
+      "eval_accuracy": 0.8589,
+      "eval_f1_macro": 0.7871987440671023,
+      "eval_f1_micro": 0.8589,
+      "eval_loss": 0.5366827845573425,
+      "eval_runtime": 25.9643,
+      "eval_samples_per_second": 385.144,
+      "eval_steps_per_second": 6.047,
+      "step": 1600
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 23.33180046081543,
+      "learning_rate": 2.112651030561479e-06,
+      "loss": 0.2409,
+      "step": 1625
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 18.71114730834961,
+      "learning_rate": 2.068230277185501e-06,
+      "loss": 0.224,
+      "step": 1650
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 21.95819854736328,
+      "learning_rate": 2.023809523809524e-06,
+      "loss": 0.2223,
+      "step": 1675
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 27.065677642822266,
+      "learning_rate": 1.979388770433547e-06,
+      "loss": 0.2787,
+      "step": 1700
+    },
+    {
+      "epoch": 1.81,
+      "eval_accuracy": 0.8653,
+      "eval_f1_macro": 0.790261134849528,
+      "eval_f1_micro": 0.8653,
+      "eval_loss": 0.5133171677589417,
+      "eval_runtime": 25.5701,
+      "eval_samples_per_second": 391.081,
+      "eval_steps_per_second": 6.14,
+      "step": 1700
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 35.288761138916016,
+      "learning_rate": 1.9349680170575695e-06,
+      "loss": 0.2709,
+      "step": 1725
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 21.077306747436523,
+      "learning_rate": 1.8905472636815921e-06,
+      "loss": 0.2002,
+      "step": 1750
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 25.394838333129883,
+      "learning_rate": 1.846126510305615e-06,
+      "loss": 0.2461,
+      "step": 1775
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 26.597759246826172,
+      "learning_rate": 1.8017057569296375e-06,
+      "loss": 0.2357,
+      "step": 1800
+    },
+    {
+      "epoch": 1.92,
+      "eval_accuracy": 0.8669,
+      "eval_f1_macro": 0.7902403947168268,
+      "eval_f1_micro": 0.8669,
+      "eval_loss": 0.5063937306404114,
+      "eval_runtime": 25.6031,
+      "eval_samples_per_second": 390.577,
+      "eval_steps_per_second": 6.132,
+      "step": 1800
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 18.62090301513672,
+      "learning_rate": 1.7572850035536603e-06,
+      "loss": 0.2612,
+      "step": 1825
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 25.897939682006836,
+      "learning_rate": 1.7128642501776832e-06,
+      "loss": 0.2243,
+      "step": 1850
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 20.556882858276367,
+      "learning_rate": 1.668443496801706e-06,
+      "loss": 0.2078,
+      "step": 1875
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 5.211686134338379,
+      "learning_rate": 1.6240227434257286e-06,
+      "loss": 0.049,
+      "step": 1900
+    },
+    {
+      "epoch": 2.03,
+      "eval_accuracy": 0.8719,
+      "eval_f1_macro": 0.797777536741942,
+      "eval_f1_micro": 0.8719,
+      "eval_loss": 0.5344421863555908,
+      "eval_runtime": 25.9399,
+      "eval_samples_per_second": 385.506,
+      "eval_steps_per_second": 6.052,
+      "step": 1900
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 14.868837356567383,
+      "learning_rate": 1.5796019900497514e-06,
+      "loss": 0.0483,
+      "step": 1925
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 3.739365577697754,
+      "learning_rate": 1.5351812366737743e-06,
+      "loss": 0.0426,
+      "step": 1950
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 3.052903413772583,
+      "learning_rate": 1.4907604832977968e-06,
+      "loss": 0.0468,
+      "step": 1975
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 11.233345985412598,
+      "learning_rate": 1.4463397299218196e-06,
+      "loss": 0.0298,
+      "step": 2000
+    },
+    {
+      "epoch": 2.13,
+      "eval_accuracy": 0.8737,
+      "eval_f1_macro": 0.7992354841882687,
+      "eval_f1_micro": 0.8737,
+      "eval_loss": 0.5761749744415283,
+      "eval_runtime": 25.6811,
+      "eval_samples_per_second": 389.392,
+      "eval_steps_per_second": 6.113,
+      "step": 2000
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 9.836750030517578,
+      "learning_rate": 1.4019189765458423e-06,
+      "loss": 0.0306,
+      "step": 2025
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 12.054607391357422,
+      "learning_rate": 1.357498223169865e-06,
+      "loss": 0.0408,
+      "step": 2050
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 2.877735137939453,
+      "learning_rate": 1.3130774697938879e-06,
+      "loss": 0.032,
+      "step": 2075
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 18.573556900024414,
+      "learning_rate": 1.2686567164179105e-06,
+      "loss": 0.0427,
+      "step": 2100
+    },
+    {
+      "epoch": 2.24,
+      "eval_accuracy": 0.8708,
+      "eval_f1_macro": 0.7976411680340069,
+      "eval_f1_micro": 0.8708,
+      "eval_loss": 0.5961406230926514,
+      "eval_runtime": 25.6941,
+      "eval_samples_per_second": 389.194,
+      "eval_steps_per_second": 6.11,
+      "step": 2100
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 11.545409202575684,
+      "learning_rate": 1.2242359630419332e-06,
+      "loss": 0.0343,
+      "step": 2125
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 3.3840279579162598,
+      "learning_rate": 1.179815209665956e-06,
+      "loss": 0.0237,
+      "step": 2150
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 7.452319145202637,
+      "learning_rate": 1.1353944562899787e-06,
+      "loss": 0.042,
+      "step": 2175
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 7.546860694885254,
+      "learning_rate": 1.0909737029140014e-06,
+      "loss": 0.036,
+      "step": 2200
+    },
+    {
+      "epoch": 2.35,
+      "eval_accuracy": 0.8728,
+      "eval_f1_macro": 0.7987820731312831,
+      "eval_f1_micro": 0.8728,
+      "eval_loss": 0.6128308773040771,
+      "eval_runtime": 25.9603,
+      "eval_samples_per_second": 385.204,
+      "eval_steps_per_second": 6.048,
+      "step": 2200
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.8773216605186462,
+      "learning_rate": 1.0465529495380243e-06,
+      "loss": 0.0264,
+      "step": 2225
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.4390593767166138,
+      "learning_rate": 1.002132196162047e-06,
+      "loss": 0.0326,
+      "step": 2250
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 3.424440622329712,
+      "learning_rate": 9.577114427860696e-07,
+      "loss": 0.0265,
+      "step": 2275
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 13.154258728027344,
+      "learning_rate": 9.132906894100925e-07,
+      "loss": 0.0551,
+      "step": 2300
+    },
+    {
+      "epoch": 2.45,
+      "eval_accuracy": 0.8708,
+      "eval_f1_macro": 0.7975921884184456,
+      "eval_f1_micro": 0.8708,
+      "eval_loss": 0.6165248155593872,
+      "eval_runtime": 25.655,
+      "eval_samples_per_second": 389.788,
+      "eval_steps_per_second": 6.12,
+      "step": 2300
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 8.90126895904541,
+      "learning_rate": 8.688699360341152e-07,
+      "loss": 0.0359,
+      "step": 2325
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 10.456062316894531,
+      "learning_rate": 8.24449182658138e-07,
+      "loss": 0.0454,
+      "step": 2350
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 13.38987922668457,
+      "learning_rate": 7.800284292821607e-07,
+      "loss": 0.0329,
+      "step": 2375
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 6.421198844909668,
+      "learning_rate": 7.356076759061834e-07,
+      "loss": 0.0392,
+      "step": 2400
+    },
+    {
+      "epoch": 2.56,
+      "eval_accuracy": 0.8749,
+      "eval_f1_macro": 0.8038155628364919,
+      "eval_f1_micro": 0.8749,
+      "eval_loss": 0.6023103594779968,
+      "eval_runtime": 25.6716,
+      "eval_samples_per_second": 389.535,
+      "eval_steps_per_second": 6.116,
+      "step": 2400
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 16.32231330871582,
+      "learning_rate": 6.911869225302062e-07,
+      "loss": 0.0319,
+      "step": 2425
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 5.884388446807861,
+      "learning_rate": 6.467661691542289e-07,
+      "loss": 0.041,
+      "step": 2450
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 17.850648880004883,
+      "learning_rate": 6.023454157782517e-07,
+      "loss": 0.036,
+      "step": 2475
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 16.628997802734375,
+      "learning_rate": 5.579246624022743e-07,
+      "loss": 0.0364,
+      "step": 2500
+    },
+    {
+      "epoch": 2.67,
+      "eval_accuracy": 0.8729,
+      "eval_f1_macro": 0.8001251167012569,
+      "eval_f1_micro": 0.8729,
+      "eval_loss": 0.6167578101158142,
+      "eval_runtime": 25.9524,
+      "eval_samples_per_second": 385.321,
+      "eval_steps_per_second": 6.05,
+      "step": 2500
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 4.081849575042725,
+      "learning_rate": 5.135039090262971e-07,
+      "loss": 0.0418,
+      "step": 2525
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 8.027618408203125,
+      "learning_rate": 4.690831556503199e-07,
+      "loss": 0.0324,
+      "step": 2550
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 9.084144592285156,
+      "learning_rate": 4.2466240227434256e-07,
+      "loss": 0.0286,
+      "step": 2575
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 5.414234161376953,
+      "learning_rate": 3.8024164889836533e-07,
+      "loss": 0.0416,
+      "step": 2600
+    },
+    {
+      "epoch": 2.77,
+      "eval_accuracy": 0.8753,
+      "eval_f1_macro": 0.8048163846871306,
+      "eval_f1_micro": 0.8753,
+      "eval_loss": 0.6102917790412903,
+      "eval_runtime": 25.9892,
+      "eval_samples_per_second": 384.775,
+      "eval_steps_per_second": 6.041,
+      "step": 2600
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 7.557031154632568,
+      "learning_rate": 3.358208955223881e-07,
+      "loss": 0.0271,
+      "step": 2625
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 5.741875648498535,
+      "learning_rate": 2.914001421464108e-07,
+      "loss": 0.0367,
+      "step": 2650
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 12.493782997131348,
+      "learning_rate": 2.4697938877043354e-07,
+      "loss": 0.0274,
+      "step": 2675
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 2.7892916202545166,
+      "learning_rate": 2.0255863539445632e-07,
+      "loss": 0.0353,
+      "step": 2700
+    },
+    {
+      "epoch": 2.88,
+      "eval_accuracy": 0.8749,
+      "eval_f1_macro": 0.8053988442835892,
+      "eval_f1_micro": 0.8749,
+      "eval_loss": 0.6117515563964844,
+      "eval_runtime": 25.6582,
+      "eval_samples_per_second": 389.74,
+      "eval_steps_per_second": 6.119,
+      "step": 2700
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 5.350805759429932,
+      "learning_rate": 1.5813788201847903e-07,
+      "loss": 0.0261,
+      "step": 2725
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 11.953265190124512,
+      "learning_rate": 1.1371712864250178e-07,
+      "loss": 0.0377,
+      "step": 2750
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 13.371731758117676,
+      "learning_rate": 6.929637526652453e-08,
+      "loss": 0.0255,
+      "step": 2775
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 15.417765617370605,
+      "learning_rate": 2.4875621890547265e-08,
+      "loss": 0.0308,
+      "step": 2800
+    },
+    {
+      "epoch": 2.99,
+      "eval_accuracy": 0.875,
+      "eval_f1_macro": 0.805663420581269,
+      "eval_f1_micro": 0.875,
+      "eval_loss": 0.611430287361145,
+      "eval_runtime": 25.9862,
+      "eval_samples_per_second": 384.819,
+      "eval_steps_per_second": 6.042,
+      "step": 2800
+    },
+    {
+      "epoch": 3.0,
+      "step": 2814,
+      "total_flos": 1.6801018930213683e+17,
+      "train_loss": 0.42635450247932005,
+      "train_runtime": 3623.2348,
+      "train_samples_per_second": 49.679,
+      "train_steps_per_second": 0.777
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 2814,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "total_flos": 1.6801018930213683e+17,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}

Qwen/Qwen1.5_1.8B_ledgar/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f26a8af75d31668209ed94c987a4170cdf12fb1b08369f54b4ac17f6f18edc56
+size 5944