Training in progress, step 1334
Browse files- checkpoint-1334/config.json +39 -0
- checkpoint-1334/merges.txt +0 -0
- checkpoint-1334/optimizer.pt +3 -0
- checkpoint-1334/pytorch_model.bin +3 -0
- checkpoint-1334/rng_state.pth +3 -0
- checkpoint-1334/scaler.pt +3 -0
- checkpoint-1334/scheduler.pt +3 -0
- checkpoint-1334/special_tokens_map.json +6 -0
- checkpoint-1334/tokenizer.json +0 -0
- checkpoint-1334/tokenizer_config.json +10 -0
- checkpoint-1334/trainer_state.json +425 -0
- checkpoint-1334/training_args.bin +3 -0
- checkpoint-1334/vocab.json +0 -0
- config.json +39 -0
- merges.txt +0 -0
- pytorch_model.bin +3 -0
- special_tokens_map.json +6 -0
- tokenizer.json +0 -0
- tokenizer_config.json +10 -0
- training_args.bin +3 -0
- vocab.json +0 -0
checkpoint-1334/config.json
ADDED
|
@@ -0,0 +1,39 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "gpt2",
|
| 3 |
+
"activation_function": "gelu_new",
|
| 4 |
+
"architectures": [
|
| 5 |
+
"GPT2LMAndValueHeadModel"
|
| 6 |
+
],
|
| 7 |
+
"attn_pdrop": 0.1,
|
| 8 |
+
"bos_token_id": 50256,
|
| 9 |
+
"embd_pdrop": 0.1,
|
| 10 |
+
"eos_token_id": 50256,
|
| 11 |
+
"initializer_range": 0.02,
|
| 12 |
+
"layer_norm_epsilon": 1e-05,
|
| 13 |
+
"model_type": "gpt2",
|
| 14 |
+
"n_ctx": 1024,
|
| 15 |
+
"n_embd": 768,
|
| 16 |
+
"n_head": 12,
|
| 17 |
+
"n_inner": null,
|
| 18 |
+
"n_layer": 12,
|
| 19 |
+
"n_positions": 1024,
|
| 20 |
+
"reorder_and_upcast_attn": true,
|
| 21 |
+
"resid_pdrop": 0.1,
|
| 22 |
+
"scale_attn_by_inverse_layer_idx": false,
|
| 23 |
+
"scale_attn_weights": true,
|
| 24 |
+
"summary_activation": null,
|
| 25 |
+
"summary_first_dropout": 0.1,
|
| 26 |
+
"summary_proj_to_labels": true,
|
| 27 |
+
"summary_type": "cls_index",
|
| 28 |
+
"summary_use_proj": true,
|
| 29 |
+
"task_specific_params": {
|
| 30 |
+
"text-generation": {
|
| 31 |
+
"do_sample": true,
|
| 32 |
+
"max_length": 50
|
| 33 |
+
}
|
| 34 |
+
},
|
| 35 |
+
"torch_dtype": "float32",
|
| 36 |
+
"transformers_version": "4.23.0",
|
| 37 |
+
"use_cache": true,
|
| 38 |
+
"vocab_size": 50257
|
| 39 |
+
}
|
checkpoint-1334/merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-1334/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:979d1530027db31f9203c7933c3402262eb346b832d52933bc497bbd9b7b1eb5
|
| 3 |
+
size 995611909
|
checkpoint-1334/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:18526a5d91cde5bfc2d5b87457cf22c4434382b1dffb1425c83b153e956c3c94
|
| 3 |
+
size 510401409
|
checkpoint-1334/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e7c2abaeca8b39188ef4d024976e6b313f1edff37e845008546436359f70d704
|
| 3 |
+
size 15533
|
checkpoint-1334/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a5970b634b76e24683de44b37ff55f568f86fe0760701f4b9ffa126dee06d439
|
| 3 |
+
size 557
|
checkpoint-1334/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1ef515473b5a52626fce1bd33c4fadf38c02a6a297eb95e59e62ff42672e042d
|
| 3 |
+
size 627
|
checkpoint-1334/special_tokens_map.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<|endoftext|>",
|
| 3 |
+
"eos_token": "<|endoftext|>",
|
| 4 |
+
"pad_token": "<|endoftext|>",
|
| 5 |
+
"unk_token": "<|endoftext|>"
|
| 6 |
+
}
|
checkpoint-1334/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-1334/tokenizer_config.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"bos_token": "<|endoftext|>",
|
| 4 |
+
"eos_token": "<|endoftext|>",
|
| 5 |
+
"model_max_length": 1024,
|
| 6 |
+
"name_or_path": "gpt2",
|
| 7 |
+
"special_tokens_map_file": null,
|
| 8 |
+
"tokenizer_class": "GPT2Tokenizer",
|
| 9 |
+
"unk_token": "<|endoftext|>"
|
| 10 |
+
}
|
checkpoint-1334/trainer_state.json
ADDED
|
@@ -0,0 +1,425 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.49962546816479403,
|
| 5 |
+
"global_step": 1334,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.0,
|
| 12 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 13 |
+
"loss": 5.8178,
|
| 14 |
+
"theoretical_loss": 10.87642657795271,
|
| 15 |
+
"tokens_seen": 1048576
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.02,
|
| 19 |
+
"learning_rate": 0.000991297767688233,
|
| 20 |
+
"loss": 3.8434,
|
| 21 |
+
"theoretical_loss": 5.240583117265738,
|
| 22 |
+
"tokens_seen": 52428800
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 0.04,
|
| 26 |
+
"learning_rate": 0.000972379871358305,
|
| 27 |
+
"loss": 3.2483,
|
| 28 |
+
"theoretical_loss": 4.741048233458233,
|
| 29 |
+
"tokens_seen": 104857600
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 0.06,
|
| 33 |
+
"learning_rate": 0.0009534619750283768,
|
| 34 |
+
"loss": 3.0804,
|
| 35 |
+
"theoretical_loss": 4.490755246681026,
|
| 36 |
+
"tokens_seen": 157286400
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 0.06,
|
| 40 |
+
"objective/train/advantage_avg": -0.008124944753944874,
|
| 41 |
+
"objective/train/docs_used": 104000,
|
| 42 |
+
"objective/train/instantaneous_batch_size": 32,
|
| 43 |
+
"objective/train/instantaneous_microbatch_size": 32768,
|
| 44 |
+
"objective/train/lm_loss": 5.698267936706543,
|
| 45 |
+
"objective/train/original_loss": 5.698267936706543,
|
| 46 |
+
"objective/train/theoretical_loss": 4.467094755136979,
|
| 47 |
+
"objective/train/tokens_used": 184300000,
|
| 48 |
+
"objective/train/value_avg": -0.0197296142578125,
|
| 49 |
+
"objective/train/value_loss": 0.00596056692302227,
|
| 50 |
+
"objective/train/value_max": -0.0023326873779296875,
|
| 51 |
+
"objective/train/value_min": -0.1529541015625,
|
| 52 |
+
"objective/train/value_reward_corr": 0.23002298967484386,
|
| 53 |
+
"objective/train/value_std": 0.01715087890625,
|
| 54 |
+
"objective/train/weight_avg": 0.9992169737815857,
|
| 55 |
+
"objective/train/weighted_lm_loss": 5.693380355834961,
|
| 56 |
+
"objective/train/weights_max": 1.01529061794281,
|
| 57 |
+
"objective/train/weights_min": 0.9343342781066895,
|
| 58 |
+
"theoretical_loss": 4.467094755136979,
|
| 59 |
+
"tokens_seen": 163840000
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.07,
|
| 63 |
+
"learning_rate": 0.0009345440786984487,
|
| 64 |
+
"loss": 2.8679,
|
| 65 |
+
"theoretical_loss": 4.3296357903425715,
|
| 66 |
+
"tokens_seen": 209715200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.09,
|
| 70 |
+
"learning_rate": 0.0009156261823685207,
|
| 71 |
+
"loss": 2.7211,
|
| 72 |
+
"theoretical_loss": 4.213299841239684,
|
| 73 |
+
"tokens_seen": 262144000
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.11,
|
| 77 |
+
"learning_rate": 0.0008967082860385925,
|
| 78 |
+
"loss": 2.5736,
|
| 79 |
+
"theoretical_loss": 4.123496734747793,
|
| 80 |
+
"tokens_seen": 314572800
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"debugging/Self-BLEU-5": 0.5365128506817183,
|
| 84 |
+
"debugging/distinct-1-grams": 0.7612814402327299,
|
| 85 |
+
"debugging/distinct-2-grams": 0.9694583753853511,
|
| 86 |
+
"debugging/entropy-1-grams": 6.003629944255698,
|
| 87 |
+
"debugging/entropy-2-grams": 7.054987089269872,
|
| 88 |
+
"debugging/length": 495.25,
|
| 89 |
+
"debugging/num_segments": 16,
|
| 90 |
+
"debugging/raw_token_scores_avg": 0.04385810345411301,
|
| 91 |
+
"debugging/raw_token_scores_std": 0.15687797963619232,
|
| 92 |
+
"epoch": 0.12,
|
| 93 |
+
"objective/train/advantage_avg": -0.026558605954051018,
|
| 94 |
+
"objective/train/docs_used": 197327,
|
| 95 |
+
"objective/train/instantaneous_batch_size": 32,
|
| 96 |
+
"objective/train/instantaneous_microbatch_size": 32768,
|
| 97 |
+
"objective/train/lm_loss": 4.860468864440918,
|
| 98 |
+
"objective/train/original_loss": 4.860468864440918,
|
| 99 |
+
"objective/train/theoretical_loss": 4.10401016644798,
|
| 100 |
+
"objective/train/tokens_used": 348140000,
|
| 101 |
+
"objective/train/value_avg": -0.01727294921875,
|
| 102 |
+
"objective/train/value_loss": 0.02569347620010376,
|
| 103 |
+
"objective/train/value_max": -0.0023593902587890625,
|
| 104 |
+
"objective/train/value_min": -0.274169921875,
|
| 105 |
+
"objective/train/value_reward_corr": -0.0460843086754045,
|
| 106 |
+
"objective/train/value_std": 0.01389312744140625,
|
| 107 |
+
"objective/train/weight_avg": 0.9974696040153503,
|
| 108 |
+
"objective/train/weighted_lm_loss": 4.853564262390137,
|
| 109 |
+
"objective/train/weights_max": 1.0277366638183594,
|
| 110 |
+
"objective/train/weights_min": 0.9056559801101685,
|
| 111 |
+
"theoretical_loss": 4.10401016644798,
|
| 112 |
+
"tokens_seen": 327680000
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 0.13,
|
| 116 |
+
"learning_rate": 0.0008777903897086645,
|
| 117 |
+
"loss": 2.4856,
|
| 118 |
+
"theoretical_loss": 4.051065245936996,
|
| 119 |
+
"tokens_seen": 367001600
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 0.15,
|
| 123 |
+
"learning_rate": 0.0008588724933787363,
|
| 124 |
+
"loss": 2.4156,
|
| 125 |
+
"theoretical_loss": 3.9908001978004064,
|
| 126 |
+
"tokens_seen": 419430400
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 0.17,
|
| 130 |
+
"learning_rate": 0.0008399545970488081,
|
| 131 |
+
"loss": 2.3438,
|
| 132 |
+
"theoretical_loss": 3.939481097700623,
|
| 133 |
+
"tokens_seen": 471859200
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 0.18,
|
| 137 |
+
"objective/train/advantage_avg": 0.01378590613603592,
|
| 138 |
+
"objective/train/docs_used": 287192,
|
| 139 |
+
"objective/train/instantaneous_batch_size": 32,
|
| 140 |
+
"objective/train/instantaneous_microbatch_size": 32768,
|
| 141 |
+
"objective/train/lm_loss": 4.726146221160889,
|
| 142 |
+
"objective/train/original_loss": 4.726146697998047,
|
| 143 |
+
"objective/train/theoretical_loss": 3.9220858822757396,
|
| 144 |
+
"objective/train/tokens_used": 511980000,
|
| 145 |
+
"objective/train/value_avg": -0.0184326171875,
|
| 146 |
+
"objective/train/value_loss": 0.0008915589423850179,
|
| 147 |
+
"objective/train/value_max": -0.0024433135986328125,
|
| 148 |
+
"objective/train/value_min": -0.169189453125,
|
| 149 |
+
"objective/train/value_reward_corr": 0.034202289497960975,
|
| 150 |
+
"objective/train/value_std": 0.01378631591796875,
|
| 151 |
+
"objective/train/weight_avg": 1.0013829469680786,
|
| 152 |
+
"objective/train/weighted_lm_loss": 4.732492446899414,
|
| 153 |
+
"objective/train/weights_max": 1.0164865255355835,
|
| 154 |
+
"objective/train/weights_min": 0.913129985332489,
|
| 155 |
+
"theoretical_loss": 3.9220858822757396,
|
| 156 |
+
"tokens_seen": 491520000
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.19,
|
| 160 |
+
"learning_rate": 0.0008210367007188801,
|
| 161 |
+
"loss": 2.2548,
|
| 162 |
+
"theoretical_loss": 3.8949869551339704,
|
| 163 |
+
"tokens_seen": 524288000
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.21,
|
| 167 |
+
"learning_rate": 0.000802118804388952,
|
| 168 |
+
"loss": 2.197,
|
| 169 |
+
"theoretical_loss": 3.855852403938689,
|
| 170 |
+
"tokens_seen": 576716800
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.22,
|
| 174 |
+
"learning_rate": 0.0007832009080590239,
|
| 175 |
+
"loss": 2.1036,
|
| 176 |
+
"theoretical_loss": 3.8210259233045254,
|
| 177 |
+
"tokens_seen": 629145600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"debugging/Self-BLEU-5": 0.5265375629586004,
|
| 181 |
+
"debugging/distinct-1-grams": 0.7435820408094715,
|
| 182 |
+
"debugging/distinct-2-grams": 0.9558103821233092,
|
| 183 |
+
"debugging/entropy-1-grams": 5.931434510687563,
|
| 184 |
+
"debugging/entropy-2-grams": 6.886416755326388,
|
| 185 |
+
"debugging/length": 521.9230769230769,
|
| 186 |
+
"debugging/num_segments": 13,
|
| 187 |
+
"debugging/raw_token_scores_avg": 0.022742915898561478,
|
| 188 |
+
"debugging/raw_token_scores_std": 0.07841178774833679,
|
| 189 |
+
"epoch": 0.23,
|
| 190 |
+
"objective/train/advantage_avg": 0.008544832468032837,
|
| 191 |
+
"objective/train/docs_used": 379091,
|
| 192 |
+
"objective/train/instantaneous_batch_size": 32,
|
| 193 |
+
"objective/train/instantaneous_microbatch_size": 32768,
|
| 194 |
+
"objective/train/lm_loss": 4.2809929847717285,
|
| 195 |
+
"objective/train/original_loss": 4.280993461608887,
|
| 196 |
+
"objective/train/theoretical_loss": 3.804976960695429,
|
| 197 |
+
"objective/train/tokens_used": 675820000,
|
| 198 |
+
"objective/train/value_avg": -0.031280517578125,
|
| 199 |
+
"objective/train/value_loss": 0.004552943632006645,
|
| 200 |
+
"objective/train/value_max": -0.0027370452880859375,
|
| 201 |
+
"objective/train/value_min": -0.306396484375,
|
| 202 |
+
"objective/train/value_reward_corr": 0.5422745268556184,
|
| 203 |
+
"objective/train/value_std": 0.03082275390625,
|
| 204 |
+
"objective/train/weight_avg": 1.000877022743225,
|
| 205 |
+
"objective/train/weighted_lm_loss": 4.284684181213379,
|
| 206 |
+
"objective/train/weights_max": 1.026648759841919,
|
| 207 |
+
"objective/train/weights_min": 0.9131191968917847,
|
| 208 |
+
"theoretical_loss": 3.804976960695429,
|
| 209 |
+
"tokens_seen": 655360000
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.24,
|
| 213 |
+
"learning_rate": 0.0007642830117290957,
|
| 214 |
+
"loss": 2.0198,
|
| 215 |
+
"theoretical_loss": 3.7897293654583164,
|
| 216 |
+
"tokens_seen": 681574400
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 0.26,
|
| 220 |
+
"learning_rate": 0.0007453651153991677,
|
| 221 |
+
"loss": 1.9114,
|
| 222 |
+
"theoretical_loss": 3.7613719997526367,
|
| 223 |
+
"tokens_seen": 734003200
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 0.28,
|
| 227 |
+
"learning_rate": 0.0007264472190692395,
|
| 228 |
+
"loss": 1.8379,
|
| 229 |
+
"theoretical_loss": 3.735495625147548,
|
| 230 |
+
"tokens_seen": 786432000
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.29,
|
| 234 |
+
"objective/train/advantage_avg": 0.005197666119784117,
|
| 235 |
+
"objective/train/docs_used": 471128,
|
| 236 |
+
"objective/train/instantaneous_batch_size": 32,
|
| 237 |
+
"objective/train/instantaneous_microbatch_size": 32768,
|
| 238 |
+
"objective/train/lm_loss": 3.469724655151367,
|
| 239 |
+
"objective/train/original_loss": 3.469724655151367,
|
| 240 |
+
"objective/train/theoretical_loss": 3.7204187214233073,
|
| 241 |
+
"objective/train/tokens_used": 839660000,
|
| 242 |
+
"objective/train/value_avg": -0.01427459716796875,
|
| 243 |
+
"objective/train/value_loss": 0.0032349335961043835,
|
| 244 |
+
"objective/train/value_max": -0.0009217262268066406,
|
| 245 |
+
"objective/train/value_min": -0.329833984375,
|
| 246 |
+
"objective/train/value_reward_corr": 0.23257723024354376,
|
| 247 |
+
"objective/train/value_std": 0.011383056640625,
|
| 248 |
+
"objective/train/weight_avg": 1.000535488128662,
|
| 249 |
+
"objective/train/weighted_lm_loss": 3.472010374069214,
|
| 250 |
+
"objective/train/weights_max": 1.0184398889541626,
|
| 251 |
+
"objective/train/weights_min": 0.9076024293899536,
|
| 252 |
+
"theoretical_loss": 3.7204187214233073,
|
| 253 |
+
"tokens_seen": 819200000
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 0.3,
|
| 257 |
+
"learning_rate": 0.0007075293227393113,
|
| 258 |
+
"loss": 1.7735,
|
| 259 |
+
"theoretical_loss": 3.7117382474521436,
|
| 260 |
+
"tokens_seen": 838860800
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.32,
|
| 264 |
+
"learning_rate": 0.0006886114264093834,
|
| 265 |
+
"loss": 1.7242,
|
| 266 |
+
"theoretical_loss": 3.689809300987042,
|
| 267 |
+
"tokens_seen": 891289600
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.34,
|
| 271 |
+
"learning_rate": 0.0006696935300794552,
|
| 272 |
+
"loss": 1.6889,
|
| 273 |
+
"theoretical_loss": 3.6694722975957066,
|
| 274 |
+
"tokens_seen": 943718400
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"debugging/Self-BLEU-5": 0.4286046663919377,
|
| 278 |
+
"debugging/distinct-1-grams": 0.8147567798871364,
|
| 279 |
+
"debugging/distinct-2-grams": 0.9823269374342457,
|
| 280 |
+
"debugging/entropy-1-grams": 6.1671920556004824,
|
| 281 |
+
"debugging/entropy-2-grams": 6.947028138756313,
|
| 282 |
+
"debugging/length": 477.53333333333336,
|
| 283 |
+
"debugging/num_segments": 15,
|
| 284 |
+
"debugging/raw_token_scores_avg": 0.020611366257071495,
|
| 285 |
+
"debugging/raw_token_scores_std": 0.08496682345867157,
|
| 286 |
+
"epoch": 0.35,
|
| 287 |
+
"objective/train/advantage_avg": -0.002937063341960311,
|
| 288 |
+
"objective/train/docs_used": 560408,
|
| 289 |
+
"objective/train/instantaneous_batch_size": 32,
|
| 290 |
+
"objective/train/instantaneous_microbatch_size": 32768,
|
| 291 |
+
"objective/train/lm_loss": 3.587904214859009,
|
| 292 |
+
"objective/train/original_loss": 3.5879039764404297,
|
| 293 |
+
"objective/train/theoretical_loss": 3.6551457544283386,
|
| 294 |
+
"objective/train/tokens_used": 1003500000,
|
| 295 |
+
"objective/train/value_avg": -0.0176849365234375,
|
| 296 |
+
"objective/train/value_loss": 0.006396747659891844,
|
| 297 |
+
"objective/train/value_max": -0.0007319450378417969,
|
| 298 |
+
"objective/train/value_min": -0.62451171875,
|
| 299 |
+
"objective/train/value_reward_corr": 0.3547212443715962,
|
| 300 |
+
"objective/train/value_std": 0.0216827392578125,
|
| 301 |
+
"objective/train/weight_avg": 0.9997376203536987,
|
| 302 |
+
"objective/train/weighted_lm_loss": 3.5868186950683594,
|
| 303 |
+
"objective/train/weights_max": 1.0237751007080078,
|
| 304 |
+
"objective/train/weights_min": 0.9077049493789673,
|
| 305 |
+
"theoretical_loss": 3.6551457544283386,
|
| 306 |
+
"tokens_seen": 983040000
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 0.36,
|
| 310 |
+
"learning_rate": 0.0006507756337495271,
|
| 311 |
+
"loss": 1.7056,
|
| 312 |
+
"theoretical_loss": 3.6505323968108674,
|
| 313 |
+
"tokens_seen": 996147200
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 0.37,
|
| 317 |
+
"learning_rate": 0.000631857737419599,
|
| 318 |
+
"loss": 1.6782,
|
| 319 |
+
"theoretical_loss": 3.632827321456789,
|
| 320 |
+
"tokens_seen": 1048576000
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.39,
|
| 324 |
+
"learning_rate": 0.0006129398410896708,
|
| 325 |
+
"loss": 1.6393,
|
| 326 |
+
"theoretical_loss": 3.616220599546101,
|
| 327 |
+
"tokens_seen": 1101004800
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.41,
|
| 331 |
+
"objective/train/advantage_avg": -0.004092915914952755,
|
| 332 |
+
"objective/train/docs_used": 649861,
|
| 333 |
+
"objective/train/instantaneous_batch_size": 32,
|
| 334 |
+
"objective/train/instantaneous_microbatch_size": 32768,
|
| 335 |
+
"objective/train/lm_loss": 3.148904323577881,
|
| 336 |
+
"objective/train/original_loss": 3.148904323577881,
|
| 337 |
+
"objective/train/theoretical_loss": 3.6024992663141386,
|
| 338 |
+
"objective/train/tokens_used": 1167340000,
|
| 339 |
+
"objective/train/value_avg": -0.0187835693359375,
|
| 340 |
+
"objective/train/value_loss": 0.007057450246065855,
|
| 341 |
+
"objective/train/value_max": -0.0004076957702636719,
|
| 342 |
+
"objective/train/value_min": -0.76025390625,
|
| 343 |
+
"objective/train/value_reward_corr": 0.5218380949829392,
|
| 344 |
+
"objective/train/value_std": 0.02960205078125,
|
| 345 |
+
"objective/train/weight_avg": 0.9996253848075867,
|
| 346 |
+
"objective/train/weighted_lm_loss": 3.1456782817840576,
|
| 347 |
+
"objective/train/weights_max": 1.0473830699920654,
|
| 348 |
+
"objective/train/weights_min": 0.9076167345046997,
|
| 349 |
+
"theoretical_loss": 3.6024992663141386,
|
| 350 |
+
"tokens_seen": 1146880000
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.41,
|
| 354 |
+
"learning_rate": 0.0005940219447597427,
|
| 355 |
+
"loss": 1.6315,
|
| 356 |
+
"theoretical_loss": 3.6005964566275575,
|
| 357 |
+
"tokens_seen": 1153433600
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.43,
|
| 361 |
+
"learning_rate": 0.0005751040484298145,
|
| 362 |
+
"loss": 1.6112,
|
| 363 |
+
"theoretical_loss": 3.585855900316411,
|
| 364 |
+
"tokens_seen": 1205862400
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 0.45,
|
| 368 |
+
"learning_rate": 0.0005561861520998866,
|
| 369 |
+
"loss": 1.5989,
|
| 370 |
+
"theoretical_loss": 3.571913680014217,
|
| 371 |
+
"tokens_seen": 1258291200
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"debugging/Self-BLEU-5": 0.49020908264157476,
|
| 375 |
+
"debugging/distinct-1-grams": 0.768901113497886,
|
| 376 |
+
"debugging/distinct-2-grams": 0.9428782333551957,
|
| 377 |
+
"debugging/entropy-1-grams": 6.085999550681761,
|
| 378 |
+
"debugging/entropy-2-grams": 7.0033060167714964,
|
| 379 |
+
"debugging/length": 490.2352941176471,
|
| 380 |
+
"debugging/num_segments": 17,
|
| 381 |
+
"debugging/raw_token_scores_avg": 0.02056093141436577,
|
| 382 |
+
"debugging/raw_token_scores_std": 0.10981010645627975,
|
| 383 |
+
"epoch": 0.47,
|
| 384 |
+
"objective/train/advantage_avg": -0.0010009908583015203,
|
| 385 |
+
"objective/train/docs_used": 741674,
|
| 386 |
+
"objective/train/instantaneous_batch_size": 32,
|
| 387 |
+
"objective/train/instantaneous_microbatch_size": 32768,
|
| 388 |
+
"objective/train/lm_loss": 3.083587408065796,
|
| 389 |
+
"objective/train/original_loss": 3.083587646484375,
|
| 390 |
+
"objective/train/theoretical_loss": 3.5586958985729016,
|
| 391 |
+
"objective/train/tokens_used": 1331180000,
|
| 392 |
+
"objective/train/value_avg": -0.0195770263671875,
|
| 393 |
+
"objective/train/value_loss": 0.007854425348341465,
|
| 394 |
+
"objective/train/value_max": -0.0004239082336425781,
|
| 395 |
+
"objective/train/value_min": -0.97705078125,
|
| 396 |
+
"objective/train/value_reward_corr": 0.5932239490084845,
|
| 397 |
+
"objective/train/value_std": 0.0595703125,
|
| 398 |
+
"objective/train/weight_avg": 0.9999384880065918,
|
| 399 |
+
"objective/train/weighted_lm_loss": 3.082087516784668,
|
| 400 |
+
"objective/train/weights_max": 1.0919088125228882,
|
| 401 |
+
"objective/train/weights_min": 0.9060803055763245,
|
| 402 |
+
"theoretical_loss": 3.5586958985729016,
|
| 403 |
+
"tokens_seen": 1310720000
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.47,
|
| 407 |
+
"learning_rate": 0.0005372682557699584,
|
| 408 |
+
"loss": 1.5642,
|
| 409 |
+
"theoretical_loss": 3.5586958985729016,
|
| 410 |
+
"tokens_seen": 1310720000
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.49,
|
| 414 |
+
"learning_rate": 0.0005183503594400303,
|
| 415 |
+
"loss": 1.5616,
|
| 416 |
+
"theoretical_loss": 3.5461381161006846,
|
| 417 |
+
"tokens_seen": 1363148800
|
| 418 |
+
}
|
| 419 |
+
],
|
| 420 |
+
"max_steps": 2670,
|
| 421 |
+
"num_train_epochs": 9223372036854775807,
|
| 422 |
+
"total_flos": 7.138646384411935e+17,
|
| 423 |
+
"trial_name": null,
|
| 424 |
+
"trial_params": null
|
| 425 |
+
}
|
checkpoint-1334/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:06649e6d12b7a4cd09a0a8cebe5d368d597d5443538b90e48182d55ad8737e4e
|
| 3 |
+
size 3451
|
checkpoint-1334/vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
config.json
ADDED
|
@@ -0,0 +1,39 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "gpt2",
|
| 3 |
+
"activation_function": "gelu_new",
|
| 4 |
+
"architectures": [
|
| 5 |
+
"GPT2LMAndValueHeadModel"
|
| 6 |
+
],
|
| 7 |
+
"attn_pdrop": 0.1,
|
| 8 |
+
"bos_token_id": 50256,
|
| 9 |
+
"embd_pdrop": 0.1,
|
| 10 |
+
"eos_token_id": 50256,
|
| 11 |
+
"initializer_range": 0.02,
|
| 12 |
+
"layer_norm_epsilon": 1e-05,
|
| 13 |
+
"model_type": "gpt2",
|
| 14 |
+
"n_ctx": 1024,
|
| 15 |
+
"n_embd": 768,
|
| 16 |
+
"n_head": 12,
|
| 17 |
+
"n_inner": null,
|
| 18 |
+
"n_layer": 12,
|
| 19 |
+
"n_positions": 1024,
|
| 20 |
+
"reorder_and_upcast_attn": true,
|
| 21 |
+
"resid_pdrop": 0.1,
|
| 22 |
+
"scale_attn_by_inverse_layer_idx": false,
|
| 23 |
+
"scale_attn_weights": true,
|
| 24 |
+
"summary_activation": null,
|
| 25 |
+
"summary_first_dropout": 0.1,
|
| 26 |
+
"summary_proj_to_labels": true,
|
| 27 |
+
"summary_type": "cls_index",
|
| 28 |
+
"summary_use_proj": true,
|
| 29 |
+
"task_specific_params": {
|
| 30 |
+
"text-generation": {
|
| 31 |
+
"do_sample": true,
|
| 32 |
+
"max_length": 50
|
| 33 |
+
}
|
| 34 |
+
},
|
| 35 |
+
"torch_dtype": "float32",
|
| 36 |
+
"transformers_version": "4.23.0",
|
| 37 |
+
"use_cache": true,
|
| 38 |
+
"vocab_size": 50257
|
| 39 |
+
}
|
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:18526a5d91cde5bfc2d5b87457cf22c4434382b1dffb1425c83b153e956c3c94
|
| 3 |
+
size 510401409
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<|endoftext|>",
|
| 3 |
+
"eos_token": "<|endoftext|>",
|
| 4 |
+
"pad_token": "<|endoftext|>",
|
| 5 |
+
"unk_token": "<|endoftext|>"
|
| 6 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"bos_token": "<|endoftext|>",
|
| 4 |
+
"eos_token": "<|endoftext|>",
|
| 5 |
+
"model_max_length": 1024,
|
| 6 |
+
"name_or_path": "gpt2",
|
| 7 |
+
"special_tokens_map_file": null,
|
| 8 |
+
"tokenizer_class": "GPT2Tokenizer",
|
| 9 |
+
"unk_token": "<|endoftext|>"
|
| 10 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:06649e6d12b7a4cd09a0a8cebe5d368d597d5443538b90e48182d55ad8737e4e
|
| 3 |
+
size 3451
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|