Auto upload 2026-08-06T21:10:31.811972
Browse files- out/glu-gelu-21L_run/checkpoint-1000/config.json +35 -0
- out/glu-gelu-21L_run/checkpoint-1000/model.safetensors +3 -0
- out/glu-gelu-21L_run/checkpoint-1000/optimizer.pt +3 -0
- out/glu-gelu-21L_run/checkpoint-1000/rng_state.pth +3 -0
- out/glu-gelu-21L_run/checkpoint-1000/scheduler.pt +3 -0
- out/glu-gelu-21L_run/checkpoint-1000/tokenizer.json +0 -0
- out/glu-gelu-21L_run/checkpoint-1000/tokenizer_config.json +13 -0
- out/glu-gelu-21L_run/checkpoint-1000/trainer_state.json +424 -0
- out/glu-gelu-21L_run/checkpoint-1000/training_args.bin +3 -0
- out/glu-gelu-21L_run/checkpoint-500/config.json +35 -0
- out/glu-gelu-21L_run/checkpoint-500/model.safetensors +3 -0
- out/glu-gelu-21L_run/checkpoint-500/optimizer.pt +3 -0
- out/glu-gelu-21L_run/checkpoint-500/rng_state.pth +3 -0
- out/glu-gelu-21L_run/checkpoint-500/scheduler.pt +3 -0
- out/glu-gelu-21L_run/checkpoint-500/tokenizer.json +0 -0
- out/glu-gelu-21L_run/checkpoint-500/tokenizer_config.json +13 -0
- out/glu-gelu-21L_run/checkpoint-500/trainer_state.json +225 -0
- out/glu-gelu-21L_run/checkpoint-500/training_args.bin +3 -0
- out/glu-gelu-21L_run/config.json +35 -0
- out/glu-gelu-21L_run/model.safetensors +3 -0
- out/glu-gelu-21L_run/training_args.bin +3 -0
- out/glu-gelu-21L_run/training_log.jsonl +0 -0
- out/glu-relu-21L_run/README.md +65 -0
- out/glu-relu-21L_run/checkpoint-2000/config.json +35 -0
- out/glu-relu-21L_run/checkpoint-2000/model.safetensors +3 -0
- out/glu-relu-21L_run/checkpoint-2000/optimizer.pt +3 -0
- out/glu-relu-21L_run/checkpoint-2000/rng_state.pth +3 -0
- out/glu-relu-21L_run/checkpoint-2000/scheduler.pt +3 -0
- out/glu-relu-21L_run/checkpoint-2000/tokenizer.json +0 -0
- out/glu-relu-21L_run/checkpoint-2000/tokenizer_config.json +13 -0
- out/glu-relu-21L_run/checkpoint-2000/trainer_state.json +814 -0
- out/glu-relu-21L_run/checkpoint-2000/training_args.bin +3 -0
- out/glu-relu-21L_run/model.safetensors +1 -1
- out/glu-relu-21L_run/tokenizer.json +0 -0
- out/glu-relu-21L_run/tokenizer_config.json +13 -0
- out/glu-relu-21L_run/training_log.jsonl +0 -0
out/glu-gelu-21L_run/checkpoint-1000/config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation": "gelu",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"TinyLlamaForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_bias": false,
|
| 7 |
+
"attention_dropout": 0.0,
|
| 8 |
+
"bos_token_id": 1,
|
| 9 |
+
"dtype": "bfloat16",
|
| 10 |
+
"eos_token_id": 2,
|
| 11 |
+
"head_dim": 32,
|
| 12 |
+
"hidden_act": "silu",
|
| 13 |
+
"hidden_size": 128,
|
| 14 |
+
"initializer_range": 0.02,
|
| 15 |
+
"intermediate_size": 256,
|
| 16 |
+
"max_position_embeddings": 512,
|
| 17 |
+
"mlp_bias": false,
|
| 18 |
+
"mlp_type": "glu",
|
| 19 |
+
"model_type": "tiny_llama",
|
| 20 |
+
"num_attention_heads": 4,
|
| 21 |
+
"num_hidden_layers": 21,
|
| 22 |
+
"num_key_value_heads": 4,
|
| 23 |
+
"pad_token_id": 0,
|
| 24 |
+
"pretraining_tp": 1,
|
| 25 |
+
"rms_norm_eps": 1e-06,
|
| 26 |
+
"rope_parameters": {
|
| 27 |
+
"rope_theta": 10000.0,
|
| 28 |
+
"rope_type": "default"
|
| 29 |
+
},
|
| 30 |
+
"tie_word_embeddings": true,
|
| 31 |
+
"tokenizer_name": "w-ahmad/tiny-stories-tokenizer",
|
| 32 |
+
"transformers_version": "5.15.0.dev0",
|
| 33 |
+
"use_cache": false,
|
| 34 |
+
"vocab_size": 4096
|
| 35 |
+
}
|
out/glu-gelu-21L_run/checkpoint-1000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ceb96e391c1617d360c574fea7b3f7d3befb41922460de233a79e982abad403f
|
| 3 |
+
size 7961528
|
out/glu-gelu-21L_run/checkpoint-1000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3a00d686eedf696ab3b9ce47a2a276e7897ff55477fe2c7eb223284c61c7f746
|
| 3 |
+
size 16041274
|
out/glu-gelu-21L_run/checkpoint-1000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:95b6047bd8cc6f4cdf7c46dea47edb8e542435510070c6cd1e0a7d9ccf5fd7da
|
| 3 |
+
size 14244
|
out/glu-gelu-21L_run/checkpoint-1000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d5f71c5fa69d56807b14256b78700be6088fcb1bda0ff8e0ea6f8709a7df2fde
|
| 3 |
+
size 1064
|
out/glu-gelu-21L_run/checkpoint-1000/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/glu-gelu-21L_run/checkpoint-1000/tokenizer_config.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": "<|endoftext|>",
|
| 5 |
+
"eos_token": "<|endoftext|>",
|
| 6 |
+
"errors": "replace",
|
| 7 |
+
"is_local": false,
|
| 8 |
+
"local_files_only": false,
|
| 9 |
+
"model_max_length": 1024,
|
| 10 |
+
"pad_token": "<|endoftext|>",
|
| 11 |
+
"tokenizer_class": "GPT2Tokenizer",
|
| 12 |
+
"unk_token": "<|endoftext|>"
|
| 13 |
+
}
|
out/glu-gelu-21L_run/checkpoint-1000/trainer_state.json
ADDED
|
@@ -0,0 +1,424 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.06740815638692282,
|
| 6 |
+
"eval_steps": 200,
|
| 7 |
+
"global_step": 1000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0013481631277384564,
|
| 14 |
+
"grad_norm": 1.2578125,
|
| 15 |
+
"learning_rate": 0.0003,
|
| 16 |
+
"loss": 7.81409912109375,
|
| 17 |
+
"step": 20
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.002696326255476913,
|
| 21 |
+
"grad_norm": 1.1796875,
|
| 22 |
+
"learning_rate": 0.0003,
|
| 23 |
+
"loss": 7.0459342956542965,
|
| 24 |
+
"step": 40
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.004044489383215369,
|
| 28 |
+
"grad_norm": 0.984375,
|
| 29 |
+
"learning_rate": 0.0003,
|
| 30 |
+
"loss": 6.482644653320312,
|
| 31 |
+
"step": 60
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.005392652510953826,
|
| 35 |
+
"grad_norm": 0.87890625,
|
| 36 |
+
"learning_rate": 0.0003,
|
| 37 |
+
"loss": 6.108085250854492,
|
| 38 |
+
"step": 80
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.006740815638692282,
|
| 42 |
+
"grad_norm": 1.171875,
|
| 43 |
+
"learning_rate": 0.0003,
|
| 44 |
+
"loss": 5.758380889892578,
|
| 45 |
+
"step": 100
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.008088978766430738,
|
| 49 |
+
"grad_norm": 1.9921875,
|
| 50 |
+
"learning_rate": 0.0003,
|
| 51 |
+
"loss": 5.2672981262207035,
|
| 52 |
+
"step": 120
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.009437141894169195,
|
| 56 |
+
"grad_norm": 2.453125,
|
| 57 |
+
"learning_rate": 0.0003,
|
| 58 |
+
"loss": 4.756313323974609,
|
| 59 |
+
"step": 140
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.010785305021907651,
|
| 63 |
+
"grad_norm": 2.515625,
|
| 64 |
+
"learning_rate": 0.0003,
|
| 65 |
+
"loss": 4.285708236694336,
|
| 66 |
+
"step": 160
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.012133468149646108,
|
| 70 |
+
"grad_norm": 2.875,
|
| 71 |
+
"learning_rate": 0.0003,
|
| 72 |
+
"loss": 3.859221267700195,
|
| 73 |
+
"step": 180
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.013481631277384564,
|
| 77 |
+
"grad_norm": 2.53125,
|
| 78 |
+
"learning_rate": 0.0003,
|
| 79 |
+
"loss": 3.458272171020508,
|
| 80 |
+
"step": 200
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.013481631277384564,
|
| 84 |
+
"eval_loss": 3.2623865604400635,
|
| 85 |
+
"eval_runtime": 9.4816,
|
| 86 |
+
"eval_samples_per_second": 1004.784,
|
| 87 |
+
"eval_steps_per_second": 15.715,
|
| 88 |
+
"step": 200
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 0.01482979440512302,
|
| 92 |
+
"grad_norm": 1.859375,
|
| 93 |
+
"learning_rate": 0.0003,
|
| 94 |
+
"loss": 3.095978546142578,
|
| 95 |
+
"step": 220
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 0.016177957532861477,
|
| 99 |
+
"grad_norm": 2.078125,
|
| 100 |
+
"learning_rate": 0.0003,
|
| 101 |
+
"loss": 2.7576486587524416,
|
| 102 |
+
"step": 240
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 0.01752612066059993,
|
| 106 |
+
"grad_norm": 1.8359375,
|
| 107 |
+
"learning_rate": 0.0003,
|
| 108 |
+
"loss": 2.4438772201538086,
|
| 109 |
+
"step": 260
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.01887428378833839,
|
| 113 |
+
"grad_norm": 2.015625,
|
| 114 |
+
"learning_rate": 0.0003,
|
| 115 |
+
"loss": 2.125438117980957,
|
| 116 |
+
"step": 280
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.020222446916076844,
|
| 120 |
+
"grad_norm": 2.046875,
|
| 121 |
+
"learning_rate": 0.0003,
|
| 122 |
+
"loss": 1.8358552932739258,
|
| 123 |
+
"step": 300
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.021570610043815303,
|
| 127 |
+
"grad_norm": 2.765625,
|
| 128 |
+
"learning_rate": 0.0003,
|
| 129 |
+
"loss": 1.5858869552612305,
|
| 130 |
+
"step": 320
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.022918773171553757,
|
| 134 |
+
"grad_norm": 1.3125,
|
| 135 |
+
"learning_rate": 0.0003,
|
| 136 |
+
"loss": 1.379590129852295,
|
| 137 |
+
"step": 340
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.024266936299292215,
|
| 141 |
+
"grad_norm": 1.4765625,
|
| 142 |
+
"learning_rate": 0.0003,
|
| 143 |
+
"loss": 1.2087390899658204,
|
| 144 |
+
"step": 360
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.02561509942703067,
|
| 148 |
+
"grad_norm": 1.484375,
|
| 149 |
+
"learning_rate": 0.0003,
|
| 150 |
+
"loss": 1.0616329193115235,
|
| 151 |
+
"step": 380
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 0.026963262554769128,
|
| 155 |
+
"grad_norm": 0.8125,
|
| 156 |
+
"learning_rate": 0.0003,
|
| 157 |
+
"loss": 0.9362197875976562,
|
| 158 |
+
"step": 400
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.026963262554769128,
|
| 162 |
+
"eval_loss": 0.8810607194900513,
|
| 163 |
+
"eval_runtime": 9.6047,
|
| 164 |
+
"eval_samples_per_second": 991.912,
|
| 165 |
+
"eval_steps_per_second": 15.513,
|
| 166 |
+
"step": 400
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.028311425682507583,
|
| 170 |
+
"grad_norm": 0.578125,
|
| 171 |
+
"learning_rate": 0.0003,
|
| 172 |
+
"loss": 0.816765308380127,
|
| 173 |
+
"step": 420
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.02965958881024604,
|
| 177 |
+
"grad_norm": 0.453125,
|
| 178 |
+
"learning_rate": 0.0003,
|
| 179 |
+
"loss": 0.7336906433105469,
|
| 180 |
+
"step": 440
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 0.031007751937984496,
|
| 184 |
+
"grad_norm": 0.4140625,
|
| 185 |
+
"learning_rate": 0.0003,
|
| 186 |
+
"loss": 0.6538865089416503,
|
| 187 |
+
"step": 460
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 0.032355915065722954,
|
| 191 |
+
"grad_norm": 0.451171875,
|
| 192 |
+
"learning_rate": 0.0003,
|
| 193 |
+
"loss": 0.603571081161499,
|
| 194 |
+
"step": 480
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.03370407819346141,
|
| 198 |
+
"grad_norm": 0.361328125,
|
| 199 |
+
"learning_rate": 0.0003,
|
| 200 |
+
"loss": 0.5583890438079834,
|
| 201 |
+
"step": 500
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.03505224132119986,
|
| 205 |
+
"grad_norm": 0.388671875,
|
| 206 |
+
"learning_rate": 0.0003,
|
| 207 |
+
"loss": 0.5258833408355713,
|
| 208 |
+
"step": 520
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.03640040444893832,
|
| 212 |
+
"grad_norm": 0.431640625,
|
| 213 |
+
"learning_rate": 0.0003,
|
| 214 |
+
"loss": 0.49422664642333985,
|
| 215 |
+
"step": 540
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.03774856757667678,
|
| 219 |
+
"grad_norm": 1.3203125,
|
| 220 |
+
"learning_rate": 0.0003,
|
| 221 |
+
"loss": 0.47668633460998533,
|
| 222 |
+
"step": 560
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 0.03909673070441524,
|
| 226 |
+
"grad_norm": 1.0390625,
|
| 227 |
+
"learning_rate": 0.0003,
|
| 228 |
+
"loss": 0.4850795269012451,
|
| 229 |
+
"step": 580
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 0.04044489383215369,
|
| 233 |
+
"grad_norm": 0.4609375,
|
| 234 |
+
"learning_rate": 0.0003,
|
| 235 |
+
"loss": 0.4641152858734131,
|
| 236 |
+
"step": 600
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.04044489383215369,
|
| 240 |
+
"eval_loss": 0.4536384344100952,
|
| 241 |
+
"eval_runtime": 9.5818,
|
| 242 |
+
"eval_samples_per_second": 994.276,
|
| 243 |
+
"eval_steps_per_second": 15.55,
|
| 244 |
+
"step": 600
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.04179305695989215,
|
| 248 |
+
"grad_norm": 0.3125,
|
| 249 |
+
"learning_rate": 0.0003,
|
| 250 |
+
"loss": 0.4424262523651123,
|
| 251 |
+
"step": 620
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.043141220087630605,
|
| 255 |
+
"grad_norm": 0.296875,
|
| 256 |
+
"learning_rate": 0.0003,
|
| 257 |
+
"loss": 0.4308023929595947,
|
| 258 |
+
"step": 640
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 0.044489383215369056,
|
| 262 |
+
"grad_norm": 0.296875,
|
| 263 |
+
"learning_rate": 0.0003,
|
| 264 |
+
"loss": 0.4142156600952148,
|
| 265 |
+
"step": 660
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.045837546343107514,
|
| 269 |
+
"grad_norm": 0.3046875,
|
| 270 |
+
"learning_rate": 0.0003,
|
| 271 |
+
"loss": 0.4043867588043213,
|
| 272 |
+
"step": 680
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.04718570947084597,
|
| 276 |
+
"grad_norm": 0.298828125,
|
| 277 |
+
"learning_rate": 0.0003,
|
| 278 |
+
"loss": 0.39342150688171384,
|
| 279 |
+
"step": 700
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.04853387259858443,
|
| 283 |
+
"grad_norm": 0.291015625,
|
| 284 |
+
"learning_rate": 0.0003,
|
| 285 |
+
"loss": 0.3856571435928345,
|
| 286 |
+
"step": 720
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 0.04988203572632288,
|
| 290 |
+
"grad_norm": 0.310546875,
|
| 291 |
+
"learning_rate": 0.0003,
|
| 292 |
+
"loss": 0.3745865821838379,
|
| 293 |
+
"step": 740
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 0.05123019885406134,
|
| 297 |
+
"grad_norm": 0.3125,
|
| 298 |
+
"learning_rate": 0.0003,
|
| 299 |
+
"loss": 0.37081801891326904,
|
| 300 |
+
"step": 760
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 0.0525783619817998,
|
| 304 |
+
"grad_norm": 0.310546875,
|
| 305 |
+
"learning_rate": 0.0003,
|
| 306 |
+
"loss": 0.3641080617904663,
|
| 307 |
+
"step": 780
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 0.053926525109538256,
|
| 311 |
+
"grad_norm": 0.314453125,
|
| 312 |
+
"learning_rate": 0.0003,
|
| 313 |
+
"loss": 0.35416674613952637,
|
| 314 |
+
"step": 800
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.053926525109538256,
|
| 318 |
+
"eval_loss": 0.35497066378593445,
|
| 319 |
+
"eval_runtime": 9.5625,
|
| 320 |
+
"eval_samples_per_second": 996.283,
|
| 321 |
+
"eval_steps_per_second": 15.582,
|
| 322 |
+
"step": 800
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 0.05527468823727671,
|
| 326 |
+
"grad_norm": 0.365234375,
|
| 327 |
+
"learning_rate": 0.0003,
|
| 328 |
+
"loss": 0.3504294157028198,
|
| 329 |
+
"step": 820
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 0.056622851365015166,
|
| 333 |
+
"grad_norm": 0.3046875,
|
| 334 |
+
"learning_rate": 0.0003,
|
| 335 |
+
"loss": 0.34422662258148196,
|
| 336 |
+
"step": 840
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 0.057971014492753624,
|
| 340 |
+
"grad_norm": 0.3203125,
|
| 341 |
+
"learning_rate": 0.0003,
|
| 342 |
+
"loss": 0.33556561470031737,
|
| 343 |
+
"step": 860
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 0.05931917762049208,
|
| 347 |
+
"grad_norm": 0.34375,
|
| 348 |
+
"learning_rate": 0.0003,
|
| 349 |
+
"loss": 0.3286173343658447,
|
| 350 |
+
"step": 880
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.06066734074823053,
|
| 354 |
+
"grad_norm": 1.4609375,
|
| 355 |
+
"learning_rate": 0.0003,
|
| 356 |
+
"loss": 0.34441075325012205,
|
| 357 |
+
"step": 900
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.06201550387596899,
|
| 361 |
+
"grad_norm": 0.58984375,
|
| 362 |
+
"learning_rate": 0.0003,
|
| 363 |
+
"loss": 0.35204482078552246,
|
| 364 |
+
"step": 920
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 0.06336366700370745,
|
| 368 |
+
"grad_norm": 0.302734375,
|
| 369 |
+
"learning_rate": 0.0003,
|
| 370 |
+
"loss": 0.32716920375823977,
|
| 371 |
+
"step": 940
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.06471183013144591,
|
| 375 |
+
"grad_norm": 0.26953125,
|
| 376 |
+
"learning_rate": 0.0003,
|
| 377 |
+
"loss": 0.3177349090576172,
|
| 378 |
+
"step": 960
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.06605999325918437,
|
| 382 |
+
"grad_norm": 0.267578125,
|
| 383 |
+
"learning_rate": 0.0003,
|
| 384 |
+
"loss": 0.31341426372528075,
|
| 385 |
+
"step": 980
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 0.06740815638692282,
|
| 389 |
+
"grad_norm": 0.26953125,
|
| 390 |
+
"learning_rate": 0.0003,
|
| 391 |
+
"loss": 0.3115365982055664,
|
| 392 |
+
"step": 1000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.06740815638692282,
|
| 396 |
+
"eval_loss": 0.3102995455265045,
|
| 397 |
+
"eval_runtime": 9.6973,
|
| 398 |
+
"eval_samples_per_second": 982.439,
|
| 399 |
+
"eval_steps_per_second": 15.365,
|
| 400 |
+
"step": 1000
|
| 401 |
+
}
|
| 402 |
+
],
|
| 403 |
+
"logging_steps": 20,
|
| 404 |
+
"max_steps": 2000,
|
| 405 |
+
"num_input_tokens_seen": 0,
|
| 406 |
+
"num_train_epochs": 1,
|
| 407 |
+
"save_steps": 500,
|
| 408 |
+
"stateful_callbacks": {
|
| 409 |
+
"TrainerControl": {
|
| 410 |
+
"args": {
|
| 411 |
+
"should_epoch_stop": false,
|
| 412 |
+
"should_evaluate": false,
|
| 413 |
+
"should_log": false,
|
| 414 |
+
"should_save": true,
|
| 415 |
+
"should_training_stop": false
|
| 416 |
+
},
|
| 417 |
+
"attributes": {}
|
| 418 |
+
}
|
| 419 |
+
},
|
| 420 |
+
"total_flos": 677539479552000.0,
|
| 421 |
+
"train_batch_size": 64,
|
| 422 |
+
"trial_name": null,
|
| 423 |
+
"trial_params": null
|
| 424 |
+
}
|
out/glu-gelu-21L_run/checkpoint-1000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:df8e0c2f7d81bcd4626a475652c0693d1cdd963c91a153a111ec8627e1e6c4bb
|
| 3 |
+
size 4920
|
out/glu-gelu-21L_run/checkpoint-500/config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation": "gelu",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"TinyLlamaForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_bias": false,
|
| 7 |
+
"attention_dropout": 0.0,
|
| 8 |
+
"bos_token_id": 1,
|
| 9 |
+
"dtype": "bfloat16",
|
| 10 |
+
"eos_token_id": 2,
|
| 11 |
+
"head_dim": 32,
|
| 12 |
+
"hidden_act": "silu",
|
| 13 |
+
"hidden_size": 128,
|
| 14 |
+
"initializer_range": 0.02,
|
| 15 |
+
"intermediate_size": 256,
|
| 16 |
+
"max_position_embeddings": 512,
|
| 17 |
+
"mlp_bias": false,
|
| 18 |
+
"mlp_type": "glu",
|
| 19 |
+
"model_type": "tiny_llama",
|
| 20 |
+
"num_attention_heads": 4,
|
| 21 |
+
"num_hidden_layers": 21,
|
| 22 |
+
"num_key_value_heads": 4,
|
| 23 |
+
"pad_token_id": 0,
|
| 24 |
+
"pretraining_tp": 1,
|
| 25 |
+
"rms_norm_eps": 1e-06,
|
| 26 |
+
"rope_parameters": {
|
| 27 |
+
"rope_theta": 10000.0,
|
| 28 |
+
"rope_type": "default"
|
| 29 |
+
},
|
| 30 |
+
"tie_word_embeddings": true,
|
| 31 |
+
"tokenizer_name": "w-ahmad/tiny-stories-tokenizer",
|
| 32 |
+
"transformers_version": "5.15.0.dev0",
|
| 33 |
+
"use_cache": false,
|
| 34 |
+
"vocab_size": 4096
|
| 35 |
+
}
|
out/glu-gelu-21L_run/checkpoint-500/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e2b1df664e602db9c04be8c1ad468daf4dd6a70615c550ab36a0d66aa9f08054
|
| 3 |
+
size 7961528
|
out/glu-gelu-21L_run/checkpoint-500/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:da32fa9f1cf97558b7cbb6b2c87b0a9b8b3192a0d97149db41a55ab9208476f3
|
| 3 |
+
size 16041274
|
out/glu-gelu-21L_run/checkpoint-500/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3cf9097d4513154245c48236b6ec5137b7ee2a21c9f58f2cba798ea275c6026f
|
| 3 |
+
size 14244
|
out/glu-gelu-21L_run/checkpoint-500/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e328dcb5672eefa48692126747c226aede5ae2e575011f8f5a2b20326a61f850
|
| 3 |
+
size 1064
|
out/glu-gelu-21L_run/checkpoint-500/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/glu-gelu-21L_run/checkpoint-500/tokenizer_config.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": "<|endoftext|>",
|
| 5 |
+
"eos_token": "<|endoftext|>",
|
| 6 |
+
"errors": "replace",
|
| 7 |
+
"is_local": false,
|
| 8 |
+
"local_files_only": false,
|
| 9 |
+
"model_max_length": 1024,
|
| 10 |
+
"pad_token": "<|endoftext|>",
|
| 11 |
+
"tokenizer_class": "GPT2Tokenizer",
|
| 12 |
+
"unk_token": "<|endoftext|>"
|
| 13 |
+
}
|
out/glu-gelu-21L_run/checkpoint-500/trainer_state.json
ADDED
|
@@ -0,0 +1,225 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.03370407819346141,
|
| 6 |
+
"eval_steps": 200,
|
| 7 |
+
"global_step": 500,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0013481631277384564,
|
| 14 |
+
"grad_norm": 1.2578125,
|
| 15 |
+
"learning_rate": 0.0003,
|
| 16 |
+
"loss": 7.81409912109375,
|
| 17 |
+
"step": 20
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.002696326255476913,
|
| 21 |
+
"grad_norm": 1.1796875,
|
| 22 |
+
"learning_rate": 0.0003,
|
| 23 |
+
"loss": 7.0459342956542965,
|
| 24 |
+
"step": 40
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.004044489383215369,
|
| 28 |
+
"grad_norm": 0.984375,
|
| 29 |
+
"learning_rate": 0.0003,
|
| 30 |
+
"loss": 6.482644653320312,
|
| 31 |
+
"step": 60
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.005392652510953826,
|
| 35 |
+
"grad_norm": 0.87890625,
|
| 36 |
+
"learning_rate": 0.0003,
|
| 37 |
+
"loss": 6.108085250854492,
|
| 38 |
+
"step": 80
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.006740815638692282,
|
| 42 |
+
"grad_norm": 1.171875,
|
| 43 |
+
"learning_rate": 0.0003,
|
| 44 |
+
"loss": 5.758380889892578,
|
| 45 |
+
"step": 100
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.008088978766430738,
|
| 49 |
+
"grad_norm": 1.9921875,
|
| 50 |
+
"learning_rate": 0.0003,
|
| 51 |
+
"loss": 5.2672981262207035,
|
| 52 |
+
"step": 120
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.009437141894169195,
|
| 56 |
+
"grad_norm": 2.453125,
|
| 57 |
+
"learning_rate": 0.0003,
|
| 58 |
+
"loss": 4.756313323974609,
|
| 59 |
+
"step": 140
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.010785305021907651,
|
| 63 |
+
"grad_norm": 2.515625,
|
| 64 |
+
"learning_rate": 0.0003,
|
| 65 |
+
"loss": 4.285708236694336,
|
| 66 |
+
"step": 160
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.012133468149646108,
|
| 70 |
+
"grad_norm": 2.875,
|
| 71 |
+
"learning_rate": 0.0003,
|
| 72 |
+
"loss": 3.859221267700195,
|
| 73 |
+
"step": 180
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.013481631277384564,
|
| 77 |
+
"grad_norm": 2.53125,
|
| 78 |
+
"learning_rate": 0.0003,
|
| 79 |
+
"loss": 3.458272171020508,
|
| 80 |
+
"step": 200
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.013481631277384564,
|
| 84 |
+
"eval_loss": 3.2623865604400635,
|
| 85 |
+
"eval_runtime": 9.4816,
|
| 86 |
+
"eval_samples_per_second": 1004.784,
|
| 87 |
+
"eval_steps_per_second": 15.715,
|
| 88 |
+
"step": 200
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 0.01482979440512302,
|
| 92 |
+
"grad_norm": 1.859375,
|
| 93 |
+
"learning_rate": 0.0003,
|
| 94 |
+
"loss": 3.095978546142578,
|
| 95 |
+
"step": 220
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 0.016177957532861477,
|
| 99 |
+
"grad_norm": 2.078125,
|
| 100 |
+
"learning_rate": 0.0003,
|
| 101 |
+
"loss": 2.7576486587524416,
|
| 102 |
+
"step": 240
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 0.01752612066059993,
|
| 106 |
+
"grad_norm": 1.8359375,
|
| 107 |
+
"learning_rate": 0.0003,
|
| 108 |
+
"loss": 2.4438772201538086,
|
| 109 |
+
"step": 260
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.01887428378833839,
|
| 113 |
+
"grad_norm": 2.015625,
|
| 114 |
+
"learning_rate": 0.0003,
|
| 115 |
+
"loss": 2.125438117980957,
|
| 116 |
+
"step": 280
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.020222446916076844,
|
| 120 |
+
"grad_norm": 2.046875,
|
| 121 |
+
"learning_rate": 0.0003,
|
| 122 |
+
"loss": 1.8358552932739258,
|
| 123 |
+
"step": 300
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.021570610043815303,
|
| 127 |
+
"grad_norm": 2.765625,
|
| 128 |
+
"learning_rate": 0.0003,
|
| 129 |
+
"loss": 1.5858869552612305,
|
| 130 |
+
"step": 320
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.022918773171553757,
|
| 134 |
+
"grad_norm": 1.3125,
|
| 135 |
+
"learning_rate": 0.0003,
|
| 136 |
+
"loss": 1.379590129852295,
|
| 137 |
+
"step": 340
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.024266936299292215,
|
| 141 |
+
"grad_norm": 1.4765625,
|
| 142 |
+
"learning_rate": 0.0003,
|
| 143 |
+
"loss": 1.2087390899658204,
|
| 144 |
+
"step": 360
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.02561509942703067,
|
| 148 |
+
"grad_norm": 1.484375,
|
| 149 |
+
"learning_rate": 0.0003,
|
| 150 |
+
"loss": 1.0616329193115235,
|
| 151 |
+
"step": 380
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 0.026963262554769128,
|
| 155 |
+
"grad_norm": 0.8125,
|
| 156 |
+
"learning_rate": 0.0003,
|
| 157 |
+
"loss": 0.9362197875976562,
|
| 158 |
+
"step": 400
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.026963262554769128,
|
| 162 |
+
"eval_loss": 0.8810607194900513,
|
| 163 |
+
"eval_runtime": 9.6047,
|
| 164 |
+
"eval_samples_per_second": 991.912,
|
| 165 |
+
"eval_steps_per_second": 15.513,
|
| 166 |
+
"step": 400
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.028311425682507583,
|
| 170 |
+
"grad_norm": 0.578125,
|
| 171 |
+
"learning_rate": 0.0003,
|
| 172 |
+
"loss": 0.816765308380127,
|
| 173 |
+
"step": 420
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.02965958881024604,
|
| 177 |
+
"grad_norm": 0.453125,
|
| 178 |
+
"learning_rate": 0.0003,
|
| 179 |
+
"loss": 0.7336906433105469,
|
| 180 |
+
"step": 440
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 0.031007751937984496,
|
| 184 |
+
"grad_norm": 0.4140625,
|
| 185 |
+
"learning_rate": 0.0003,
|
| 186 |
+
"loss": 0.6538865089416503,
|
| 187 |
+
"step": 460
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 0.032355915065722954,
|
| 191 |
+
"grad_norm": 0.451171875,
|
| 192 |
+
"learning_rate": 0.0003,
|
| 193 |
+
"loss": 0.603571081161499,
|
| 194 |
+
"step": 480
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.03370407819346141,
|
| 198 |
+
"grad_norm": 0.361328125,
|
| 199 |
+
"learning_rate": 0.0003,
|
| 200 |
+
"loss": 0.5583890438079834,
|
| 201 |
+
"step": 500
|
| 202 |
+
}
|
| 203 |
+
],
|
| 204 |
+
"logging_steps": 20,
|
| 205 |
+
"max_steps": 2000,
|
| 206 |
+
"num_input_tokens_seen": 0,
|
| 207 |
+
"num_train_epochs": 1,
|
| 208 |
+
"save_steps": 500,
|
| 209 |
+
"stateful_callbacks": {
|
| 210 |
+
"TrainerControl": {
|
| 211 |
+
"args": {
|
| 212 |
+
"should_epoch_stop": false,
|
| 213 |
+
"should_evaluate": false,
|
| 214 |
+
"should_log": false,
|
| 215 |
+
"should_save": true,
|
| 216 |
+
"should_training_stop": false
|
| 217 |
+
},
|
| 218 |
+
"attributes": {}
|
| 219 |
+
}
|
| 220 |
+
},
|
| 221 |
+
"total_flos": 338769739776000.0,
|
| 222 |
+
"train_batch_size": 64,
|
| 223 |
+
"trial_name": null,
|
| 224 |
+
"trial_params": null
|
| 225 |
+
}
|
out/glu-gelu-21L_run/checkpoint-500/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:df8e0c2f7d81bcd4626a475652c0693d1cdd963c91a153a111ec8627e1e6c4bb
|
| 3 |
+
size 4920
|
out/glu-gelu-21L_run/config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation": "gelu",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"TinyLlamaForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_bias": false,
|
| 7 |
+
"attention_dropout": 0.0,
|
| 8 |
+
"bos_token_id": 1,
|
| 9 |
+
"dtype": "bfloat16",
|
| 10 |
+
"eos_token_id": 2,
|
| 11 |
+
"head_dim": 32,
|
| 12 |
+
"hidden_act": "silu",
|
| 13 |
+
"hidden_size": 128,
|
| 14 |
+
"initializer_range": 0.02,
|
| 15 |
+
"intermediate_size": 256,
|
| 16 |
+
"max_position_embeddings": 512,
|
| 17 |
+
"mlp_bias": false,
|
| 18 |
+
"mlp_type": "glu",
|
| 19 |
+
"model_type": "tiny_llama",
|
| 20 |
+
"num_attention_heads": 4,
|
| 21 |
+
"num_hidden_layers": 21,
|
| 22 |
+
"num_key_value_heads": 4,
|
| 23 |
+
"pad_token_id": 0,
|
| 24 |
+
"pretraining_tp": 1,
|
| 25 |
+
"rms_norm_eps": 1e-06,
|
| 26 |
+
"rope_parameters": {
|
| 27 |
+
"rope_theta": 10000.0,
|
| 28 |
+
"rope_type": "default"
|
| 29 |
+
},
|
| 30 |
+
"tie_word_embeddings": true,
|
| 31 |
+
"tokenizer_name": "w-ahmad/tiny-stories-tokenizer",
|
| 32 |
+
"transformers_version": "5.15.0.dev0",
|
| 33 |
+
"use_cache": false,
|
| 34 |
+
"vocab_size": 4096
|
| 35 |
+
}
|
out/glu-gelu-21L_run/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ceb96e391c1617d360c574fea7b3f7d3befb41922460de233a79e982abad403f
|
| 3 |
+
size 7961528
|
out/glu-gelu-21L_run/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:df8e0c2f7d81bcd4626a475652c0693d1cdd963c91a153a111ec8627e1e6c4bb
|
| 3 |
+
size 4920
|
out/glu-gelu-21L_run/training_log.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/glu-relu-21L_run/README.md
ADDED
|
@@ -0,0 +1,65 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
tags:
|
| 4 |
+
- generated_from_trainer
|
| 5 |
+
model-index:
|
| 6 |
+
- name: ACTio-glu-relu-21L
|
| 7 |
+
results: []
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 11 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 12 |
+
|
| 13 |
+
# ACTio-glu-relu-21L
|
| 14 |
+
|
| 15 |
+
This model is a fine-tuned version of [](https://huggingface.co/) on an unknown dataset.
|
| 16 |
+
It achieves the following results on the evaluation set:
|
| 17 |
+
- Loss: 0.2269
|
| 18 |
+
|
| 19 |
+
## Model description
|
| 20 |
+
|
| 21 |
+
More information needed
|
| 22 |
+
|
| 23 |
+
## Intended uses & limitations
|
| 24 |
+
|
| 25 |
+
More information needed
|
| 26 |
+
|
| 27 |
+
## Training and evaluation data
|
| 28 |
+
|
| 29 |
+
More information needed
|
| 30 |
+
|
| 31 |
+
## Training procedure
|
| 32 |
+
|
| 33 |
+
### Training hyperparameters
|
| 34 |
+
|
| 35 |
+
The following hyperparameters were used during training:
|
| 36 |
+
- learning_rate: 0.0003
|
| 37 |
+
- train_batch_size: 64
|
| 38 |
+
- eval_batch_size: 64
|
| 39 |
+
- seed: 42
|
| 40 |
+
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
|
| 41 |
+
- lr_scheduler_type: constant
|
| 42 |
+
- training_steps: 2000
|
| 43 |
+
|
| 44 |
+
### Training results
|
| 45 |
+
|
| 46 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
| 47 |
+
|:-------------:|:------:|:----:|:---------------:|
|
| 48 |
+
| 3.3777 | 0.0135 | 200 | 3.1845 |
|
| 49 |
+
| 0.8929 | 0.0270 | 400 | 0.8398 |
|
| 50 |
+
| 0.4342 | 0.0404 | 600 | 0.4316 |
|
| 51 |
+
| 0.3467 | 0.0539 | 800 | 0.3478 |
|
| 52 |
+
| 0.3036 | 0.0674 | 1000 | 0.3025 |
|
| 53 |
+
| 0.2874 | 0.0809 | 1200 | 0.2802 |
|
| 54 |
+
| 0.2546 | 0.0944 | 1400 | 0.2565 |
|
| 55 |
+
| 0.2542 | 0.1079 | 1600 | 0.2647 |
|
| 56 |
+
| 0.2347 | 0.1213 | 1800 | 0.2337 |
|
| 57 |
+
| 0.2253 | 0.1348 | 2000 | 0.2269 |
|
| 58 |
+
|
| 59 |
+
|
| 60 |
+
### Framework versions
|
| 61 |
+
|
| 62 |
+
- Transformers 5.15.0.dev0
|
| 63 |
+
- Pytorch 2.6.0+cu124
|
| 64 |
+
- Datasets 5.0.1
|
| 65 |
+
- Tokenizers 0.22.2
|
out/glu-relu-21L_run/checkpoint-2000/config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation": "relu",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"TinyLlamaForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_bias": false,
|
| 7 |
+
"attention_dropout": 0.0,
|
| 8 |
+
"bos_token_id": 1,
|
| 9 |
+
"dtype": "bfloat16",
|
| 10 |
+
"eos_token_id": 2,
|
| 11 |
+
"head_dim": 32,
|
| 12 |
+
"hidden_act": "silu",
|
| 13 |
+
"hidden_size": 128,
|
| 14 |
+
"initializer_range": 0.02,
|
| 15 |
+
"intermediate_size": 256,
|
| 16 |
+
"max_position_embeddings": 512,
|
| 17 |
+
"mlp_bias": false,
|
| 18 |
+
"mlp_type": "glu",
|
| 19 |
+
"model_type": "tiny_llama",
|
| 20 |
+
"num_attention_heads": 4,
|
| 21 |
+
"num_hidden_layers": 21,
|
| 22 |
+
"num_key_value_heads": 4,
|
| 23 |
+
"pad_token_id": 0,
|
| 24 |
+
"pretraining_tp": 1,
|
| 25 |
+
"rms_norm_eps": 1e-06,
|
| 26 |
+
"rope_parameters": {
|
| 27 |
+
"rope_theta": 10000.0,
|
| 28 |
+
"rope_type": "default"
|
| 29 |
+
},
|
| 30 |
+
"tie_word_embeddings": true,
|
| 31 |
+
"tokenizer_name": "w-ahmad/tiny-stories-tokenizer",
|
| 32 |
+
"transformers_version": "5.15.0.dev0",
|
| 33 |
+
"use_cache": false,
|
| 34 |
+
"vocab_size": 4096
|
| 35 |
+
}
|
out/glu-relu-21L_run/checkpoint-2000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:49014ed1c74d9604971be6d4a4d168cf9abc3c482421411cf1a7bc19b02ca114
|
| 3 |
+
size 7961528
|
out/glu-relu-21L_run/checkpoint-2000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:31afab93c7375aff5d917b1cd0107893f5384321d2da5a9cee0d1daef7de5049
|
| 3 |
+
size 16041274
|
out/glu-relu-21L_run/checkpoint-2000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b1a97db8e41139aa1239ba7fb79ddeb0af5998c6305a440c1fe182e6ad02f2f5
|
| 3 |
+
size 14244
|
out/glu-relu-21L_run/checkpoint-2000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e91519ad666af6acc4beafd44aa617aefb90f632388a0f75cbf88f957400c190
|
| 3 |
+
size 1064
|
out/glu-relu-21L_run/checkpoint-2000/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/glu-relu-21L_run/checkpoint-2000/tokenizer_config.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": "<|endoftext|>",
|
| 5 |
+
"eos_token": "<|endoftext|>",
|
| 6 |
+
"errors": "replace",
|
| 7 |
+
"is_local": false,
|
| 8 |
+
"local_files_only": false,
|
| 9 |
+
"model_max_length": 1024,
|
| 10 |
+
"pad_token": "<|endoftext|>",
|
| 11 |
+
"tokenizer_class": "GPT2Tokenizer",
|
| 12 |
+
"unk_token": "<|endoftext|>"
|
| 13 |
+
}
|
out/glu-relu-21L_run/checkpoint-2000/trainer_state.json
ADDED
|
@@ -0,0 +1,814 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.13481631277384565,
|
| 6 |
+
"eval_steps": 200,
|
| 7 |
+
"global_step": 2000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0013481631277384564,
|
| 14 |
+
"grad_norm": 1.2578125,
|
| 15 |
+
"learning_rate": 0.0003,
|
| 16 |
+
"loss": 7.8173583984375,
|
| 17 |
+
"step": 20
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.002696326255476913,
|
| 21 |
+
"grad_norm": 1.1796875,
|
| 22 |
+
"learning_rate": 0.0003,
|
| 23 |
+
"loss": 7.044731903076172,
|
| 24 |
+
"step": 40
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.004044489383215369,
|
| 28 |
+
"grad_norm": 0.9921875,
|
| 29 |
+
"learning_rate": 0.0003,
|
| 30 |
+
"loss": 6.4738929748535154,
|
| 31 |
+
"step": 60
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.005392652510953826,
|
| 35 |
+
"grad_norm": 1.109375,
|
| 36 |
+
"learning_rate": 0.0003,
|
| 37 |
+
"loss": 6.081282424926758,
|
| 38 |
+
"step": 80
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.006740815638692282,
|
| 42 |
+
"grad_norm": 2.140625,
|
| 43 |
+
"learning_rate": 0.0003,
|
| 44 |
+
"loss": 5.666643524169922,
|
| 45 |
+
"step": 100
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.008088978766430738,
|
| 49 |
+
"grad_norm": 1.8203125,
|
| 50 |
+
"learning_rate": 0.0003,
|
| 51 |
+
"loss": 5.143213272094727,
|
| 52 |
+
"step": 120
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.009437141894169195,
|
| 56 |
+
"grad_norm": 2.09375,
|
| 57 |
+
"learning_rate": 0.0003,
|
| 58 |
+
"loss": 4.639799118041992,
|
| 59 |
+
"step": 140
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.010785305021907651,
|
| 63 |
+
"grad_norm": 2.453125,
|
| 64 |
+
"learning_rate": 0.0003,
|
| 65 |
+
"loss": 4.185822677612305,
|
| 66 |
+
"step": 160
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.012133468149646108,
|
| 70 |
+
"grad_norm": 2.75,
|
| 71 |
+
"learning_rate": 0.0003,
|
| 72 |
+
"loss": 3.775371551513672,
|
| 73 |
+
"step": 180
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.013481631277384564,
|
| 77 |
+
"grad_norm": 2.703125,
|
| 78 |
+
"learning_rate": 0.0003,
|
| 79 |
+
"loss": 3.377668762207031,
|
| 80 |
+
"step": 200
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.013481631277384564,
|
| 84 |
+
"eval_loss": 3.1845014095306396,
|
| 85 |
+
"eval_runtime": 9.4304,
|
| 86 |
+
"eval_samples_per_second": 1010.238,
|
| 87 |
+
"eval_steps_per_second": 15.8,
|
| 88 |
+
"step": 200
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 0.01482979440512302,
|
| 92 |
+
"grad_norm": 2.453125,
|
| 93 |
+
"learning_rate": 0.0003,
|
| 94 |
+
"loss": 3.025987243652344,
|
| 95 |
+
"step": 220
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 0.016177957532861477,
|
| 99 |
+
"grad_norm": 2.046875,
|
| 100 |
+
"learning_rate": 0.0003,
|
| 101 |
+
"loss": 2.687646675109863,
|
| 102 |
+
"step": 240
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 0.01752612066059993,
|
| 106 |
+
"grad_norm": 2.5,
|
| 107 |
+
"learning_rate": 0.0003,
|
| 108 |
+
"loss": 2.3610048294067383,
|
| 109 |
+
"step": 260
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.01887428378833839,
|
| 113 |
+
"grad_norm": 2.09375,
|
| 114 |
+
"learning_rate": 0.0003,
|
| 115 |
+
"loss": 2.034983444213867,
|
| 116 |
+
"step": 280
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.020222446916076844,
|
| 120 |
+
"grad_norm": 1.578125,
|
| 121 |
+
"learning_rate": 0.0003,
|
| 122 |
+
"loss": 1.7438932418823243,
|
| 123 |
+
"step": 300
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.021570610043815303,
|
| 127 |
+
"grad_norm": 1.2578125,
|
| 128 |
+
"learning_rate": 0.0003,
|
| 129 |
+
"loss": 1.4917546272277833,
|
| 130 |
+
"step": 320
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.022918773171553757,
|
| 134 |
+
"grad_norm": 1.984375,
|
| 135 |
+
"learning_rate": 0.0003,
|
| 136 |
+
"loss": 1.282207202911377,
|
| 137 |
+
"step": 340
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.024266936299292215,
|
| 141 |
+
"grad_norm": 1.7890625,
|
| 142 |
+
"learning_rate": 0.0003,
|
| 143 |
+
"loss": 1.1357386589050293,
|
| 144 |
+
"step": 360
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.02561509942703067,
|
| 148 |
+
"grad_norm": 1.4140625,
|
| 149 |
+
"learning_rate": 0.0003,
|
| 150 |
+
"loss": 1.0109553337097168,
|
| 151 |
+
"step": 380
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 0.026963262554769128,
|
| 155 |
+
"grad_norm": 0.703125,
|
| 156 |
+
"learning_rate": 0.0003,
|
| 157 |
+
"loss": 0.8929083824157715,
|
| 158 |
+
"step": 400
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.026963262554769128,
|
| 162 |
+
"eval_loss": 0.8398274779319763,
|
| 163 |
+
"eval_runtime": 9.3507,
|
| 164 |
+
"eval_samples_per_second": 1018.856,
|
| 165 |
+
"eval_steps_per_second": 15.935,
|
| 166 |
+
"step": 400
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.028311425682507583,
|
| 170 |
+
"grad_norm": 0.48046875,
|
| 171 |
+
"learning_rate": 0.0003,
|
| 172 |
+
"loss": 0.7798076152801514,
|
| 173 |
+
"step": 420
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.02965958881024604,
|
| 177 |
+
"grad_norm": 0.4375,
|
| 178 |
+
"learning_rate": 0.0003,
|
| 179 |
+
"loss": 0.7058346271514893,
|
| 180 |
+
"step": 440
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 0.031007751937984496,
|
| 184 |
+
"grad_norm": 0.42578125,
|
| 185 |
+
"learning_rate": 0.0003,
|
| 186 |
+
"loss": 0.6319131851196289,
|
| 187 |
+
"step": 460
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 0.032355915065722954,
|
| 191 |
+
"grad_norm": 0.427734375,
|
| 192 |
+
"learning_rate": 0.0003,
|
| 193 |
+
"loss": 0.5855731964111328,
|
| 194 |
+
"step": 480
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.03370407819346141,
|
| 198 |
+
"grad_norm": 0.392578125,
|
| 199 |
+
"learning_rate": 0.0003,
|
| 200 |
+
"loss": 0.5416285514831543,
|
| 201 |
+
"step": 500
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.03505224132119986,
|
| 205 |
+
"grad_norm": 0.376953125,
|
| 206 |
+
"learning_rate": 0.0003,
|
| 207 |
+
"loss": 0.5094103813171387,
|
| 208 |
+
"step": 520
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.03640040444893832,
|
| 212 |
+
"grad_norm": 0.40234375,
|
| 213 |
+
"learning_rate": 0.0003,
|
| 214 |
+
"loss": 0.4790643215179443,
|
| 215 |
+
"step": 540
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.03774856757667678,
|
| 219 |
+
"grad_norm": 0.365234375,
|
| 220 |
+
"learning_rate": 0.0003,
|
| 221 |
+
"loss": 0.4618992805480957,
|
| 222 |
+
"step": 560
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 0.03909673070441524,
|
| 226 |
+
"grad_norm": 0.33984375,
|
| 227 |
+
"learning_rate": 0.0003,
|
| 228 |
+
"loss": 0.4436288356781006,
|
| 229 |
+
"step": 580
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 0.04044489383215369,
|
| 233 |
+
"grad_norm": 1.7421875,
|
| 234 |
+
"learning_rate": 0.0003,
|
| 235 |
+
"loss": 0.4342087745666504,
|
| 236 |
+
"step": 600
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.04044489383215369,
|
| 240 |
+
"eval_loss": 0.43162089586257935,
|
| 241 |
+
"eval_runtime": 9.7516,
|
| 242 |
+
"eval_samples_per_second": 976.972,
|
| 243 |
+
"eval_steps_per_second": 15.28,
|
| 244 |
+
"step": 600
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.04179305695989215,
|
| 248 |
+
"grad_norm": 1.421875,
|
| 249 |
+
"learning_rate": 0.0003,
|
| 250 |
+
"loss": 0.44816021919250487,
|
| 251 |
+
"step": 620
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.043141220087630605,
|
| 255 |
+
"grad_norm": 0.408203125,
|
| 256 |
+
"learning_rate": 0.0003,
|
| 257 |
+
"loss": 0.4273179531097412,
|
| 258 |
+
"step": 640
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 0.044489383215369056,
|
| 262 |
+
"grad_norm": 0.326171875,
|
| 263 |
+
"learning_rate": 0.0003,
|
| 264 |
+
"loss": 0.40578336715698243,
|
| 265 |
+
"step": 660
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.045837546343107514,
|
| 269 |
+
"grad_norm": 0.296875,
|
| 270 |
+
"learning_rate": 0.0003,
|
| 271 |
+
"loss": 0.3955844402313232,
|
| 272 |
+
"step": 680
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.04718570947084597,
|
| 276 |
+
"grad_norm": 0.291015625,
|
| 277 |
+
"learning_rate": 0.0003,
|
| 278 |
+
"loss": 0.3847493648529053,
|
| 279 |
+
"step": 700
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.04853387259858443,
|
| 283 |
+
"grad_norm": 0.287109375,
|
| 284 |
+
"learning_rate": 0.0003,
|
| 285 |
+
"loss": 0.3775057315826416,
|
| 286 |
+
"step": 720
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 0.04988203572632288,
|
| 290 |
+
"grad_norm": 0.294921875,
|
| 291 |
+
"learning_rate": 0.0003,
|
| 292 |
+
"loss": 0.3667799472808838,
|
| 293 |
+
"step": 740
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 0.05123019885406134,
|
| 297 |
+
"grad_norm": 0.28515625,
|
| 298 |
+
"learning_rate": 0.0003,
|
| 299 |
+
"loss": 0.36283140182495116,
|
| 300 |
+
"step": 760
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 0.0525783619817998,
|
| 304 |
+
"grad_norm": 0.306640625,
|
| 305 |
+
"learning_rate": 0.0003,
|
| 306 |
+
"loss": 0.35626342296600344,
|
| 307 |
+
"step": 780
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 0.053926525109538256,
|
| 311 |
+
"grad_norm": 0.302734375,
|
| 312 |
+
"learning_rate": 0.0003,
|
| 313 |
+
"loss": 0.34672117233276367,
|
| 314 |
+
"step": 800
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.053926525109538256,
|
| 318 |
+
"eval_loss": 0.34778550267219543,
|
| 319 |
+
"eval_runtime": 9.5613,
|
| 320 |
+
"eval_samples_per_second": 996.41,
|
| 321 |
+
"eval_steps_per_second": 15.584,
|
| 322 |
+
"step": 800
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 0.05527468823727671,
|
| 326 |
+
"grad_norm": 0.341796875,
|
| 327 |
+
"learning_rate": 0.0003,
|
| 328 |
+
"loss": 0.34303557872772217,
|
| 329 |
+
"step": 820
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 0.056622851365015166,
|
| 333 |
+
"grad_norm": 0.373046875,
|
| 334 |
+
"learning_rate": 0.0003,
|
| 335 |
+
"loss": 0.337309741973877,
|
| 336 |
+
"step": 840
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 0.057971014492753624,
|
| 340 |
+
"grad_norm": 0.302734375,
|
| 341 |
+
"learning_rate": 0.0003,
|
| 342 |
+
"loss": 0.32960782051086424,
|
| 343 |
+
"step": 860
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 0.05931917762049208,
|
| 347 |
+
"grad_norm": 0.328125,
|
| 348 |
+
"learning_rate": 0.0003,
|
| 349 |
+
"loss": 0.3231168031692505,
|
| 350 |
+
"step": 880
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.06066734074823053,
|
| 354 |
+
"grad_norm": 0.326171875,
|
| 355 |
+
"learning_rate": 0.0003,
|
| 356 |
+
"loss": 0.32200720310211184,
|
| 357 |
+
"step": 900
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.06201550387596899,
|
| 361 |
+
"grad_norm": 0.291015625,
|
| 362 |
+
"learning_rate": 0.0003,
|
| 363 |
+
"loss": 0.317995285987854,
|
| 364 |
+
"step": 920
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 0.06336366700370745,
|
| 368 |
+
"grad_norm": 0.326171875,
|
| 369 |
+
"learning_rate": 0.0003,
|
| 370 |
+
"loss": 0.3131638288497925,
|
| 371 |
+
"step": 940
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.06471183013144591,
|
| 375 |
+
"grad_norm": 0.3359375,
|
| 376 |
+
"learning_rate": 0.0003,
|
| 377 |
+
"loss": 0.30775227546691897,
|
| 378 |
+
"step": 960
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.06605999325918437,
|
| 382 |
+
"grad_norm": 0.328125,
|
| 383 |
+
"learning_rate": 0.0003,
|
| 384 |
+
"loss": 0.3050234317779541,
|
| 385 |
+
"step": 980
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 0.06740815638692282,
|
| 389 |
+
"grad_norm": 0.345703125,
|
| 390 |
+
"learning_rate": 0.0003,
|
| 391 |
+
"loss": 0.3036006450653076,
|
| 392 |
+
"step": 1000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.06740815638692282,
|
| 396 |
+
"eval_loss": 0.3025302290916443,
|
| 397 |
+
"eval_runtime": 9.5202,
|
| 398 |
+
"eval_samples_per_second": 1000.717,
|
| 399 |
+
"eval_steps_per_second": 15.651,
|
| 400 |
+
"step": 1000
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 0.06875631951466127,
|
| 404 |
+
"grad_norm": 2.859375,
|
| 405 |
+
"learning_rate": 0.0003,
|
| 406 |
+
"loss": 0.3098264694213867,
|
| 407 |
+
"step": 1020
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 0.07010448264239973,
|
| 411 |
+
"grad_norm": 0.9609375,
|
| 412 |
+
"learning_rate": 0.0003,
|
| 413 |
+
"loss": 0.3203303337097168,
|
| 414 |
+
"step": 1040
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.07145264577013818,
|
| 418 |
+
"grad_norm": 0.3515625,
|
| 419 |
+
"learning_rate": 0.0003,
|
| 420 |
+
"loss": 0.30220096111297606,
|
| 421 |
+
"step": 1060
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 0.07280080889787664,
|
| 425 |
+
"grad_norm": 0.267578125,
|
| 426 |
+
"learning_rate": 0.0003,
|
| 427 |
+
"loss": 0.2920833587646484,
|
| 428 |
+
"step": 1080
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.0741489720256151,
|
| 432 |
+
"grad_norm": 0.26953125,
|
| 433 |
+
"learning_rate": 0.0003,
|
| 434 |
+
"loss": 0.289125919342041,
|
| 435 |
+
"step": 1100
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.07549713515335356,
|
| 439 |
+
"grad_norm": 0.263671875,
|
| 440 |
+
"learning_rate": 0.0003,
|
| 441 |
+
"loss": 0.2848074913024902,
|
| 442 |
+
"step": 1120
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.07684529828109202,
|
| 446 |
+
"grad_norm": 0.25,
|
| 447 |
+
"learning_rate": 0.0003,
|
| 448 |
+
"loss": 0.27943434715271,
|
| 449 |
+
"step": 1140
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 0.07819346140883048,
|
| 453 |
+
"grad_norm": 0.255859375,
|
| 454 |
+
"learning_rate": 0.0003,
|
| 455 |
+
"loss": 0.2784644842147827,
|
| 456 |
+
"step": 1160
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 0.07954162453656892,
|
| 460 |
+
"grad_norm": 0.94921875,
|
| 461 |
+
"learning_rate": 0.0003,
|
| 462 |
+
"loss": 0.2912662744522095,
|
| 463 |
+
"step": 1180
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 0.08088978766430738,
|
| 467 |
+
"grad_norm": 0.3671875,
|
| 468 |
+
"learning_rate": 0.0003,
|
| 469 |
+
"loss": 0.28743181228637693,
|
| 470 |
+
"step": 1200
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.08088978766430738,
|
| 474 |
+
"eval_loss": 0.2801724672317505,
|
| 475 |
+
"eval_runtime": 9.569,
|
| 476 |
+
"eval_samples_per_second": 995.606,
|
| 477 |
+
"eval_steps_per_second": 15.571,
|
| 478 |
+
"step": 1200
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.08223795079204584,
|
| 482 |
+
"grad_norm": 0.2578125,
|
| 483 |
+
"learning_rate": 0.0003,
|
| 484 |
+
"loss": 0.27456440925598147,
|
| 485 |
+
"step": 1220
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.0835861139197843,
|
| 489 |
+
"grad_norm": 0.2451171875,
|
| 490 |
+
"learning_rate": 0.0003,
|
| 491 |
+
"loss": 0.271167516708374,
|
| 492 |
+
"step": 1240
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.08493427704752275,
|
| 496 |
+
"grad_norm": 0.2431640625,
|
| 497 |
+
"learning_rate": 0.0003,
|
| 498 |
+
"loss": 0.269512939453125,
|
| 499 |
+
"step": 1260
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.08628244017526121,
|
| 503 |
+
"grad_norm": 0.2431640625,
|
| 504 |
+
"learning_rate": 0.0003,
|
| 505 |
+
"loss": 0.26471068859100344,
|
| 506 |
+
"step": 1280
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.08763060330299967,
|
| 510 |
+
"grad_norm": 0.2451171875,
|
| 511 |
+
"learning_rate": 0.0003,
|
| 512 |
+
"loss": 0.2643667936325073,
|
| 513 |
+
"step": 1300
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.08897876643073811,
|
| 517 |
+
"grad_norm": 0.23828125,
|
| 518 |
+
"learning_rate": 0.0003,
|
| 519 |
+
"loss": 0.26134300231933594,
|
| 520 |
+
"step": 1320
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.09032692955847657,
|
| 524 |
+
"grad_norm": 0.240234375,
|
| 525 |
+
"learning_rate": 0.0003,
|
| 526 |
+
"loss": 0.2599949836730957,
|
| 527 |
+
"step": 1340
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.09167509268621503,
|
| 531 |
+
"grad_norm": 0.2421875,
|
| 532 |
+
"learning_rate": 0.0003,
|
| 533 |
+
"loss": 0.25742900371551514,
|
| 534 |
+
"step": 1360
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.09302325581395349,
|
| 538 |
+
"grad_norm": 0.240234375,
|
| 539 |
+
"learning_rate": 0.0003,
|
| 540 |
+
"loss": 0.2569489240646362,
|
| 541 |
+
"step": 1380
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.09437141894169195,
|
| 545 |
+
"grad_norm": 0.2431640625,
|
| 546 |
+
"learning_rate": 0.0003,
|
| 547 |
+
"loss": 0.25456163883209226,
|
| 548 |
+
"step": 1400
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.09437141894169195,
|
| 552 |
+
"eval_loss": 0.2564834654331207,
|
| 553 |
+
"eval_runtime": 9.5574,
|
| 554 |
+
"eval_samples_per_second": 996.815,
|
| 555 |
+
"eval_steps_per_second": 15.59,
|
| 556 |
+
"step": 1400
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.0957195820694304,
|
| 560 |
+
"grad_norm": 0.2431640625,
|
| 561 |
+
"learning_rate": 0.0003,
|
| 562 |
+
"loss": 0.2542454957962036,
|
| 563 |
+
"step": 1420
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.09706774519716886,
|
| 567 |
+
"grad_norm": 0.2373046875,
|
| 568 |
+
"learning_rate": 0.0003,
|
| 569 |
+
"loss": 0.25176928043365476,
|
| 570 |
+
"step": 1440
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.09841590832490732,
|
| 574 |
+
"grad_norm": 0.2431640625,
|
| 575 |
+
"learning_rate": 0.0003,
|
| 576 |
+
"loss": 0.25000886917114257,
|
| 577 |
+
"step": 1460
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.09976407145264576,
|
| 581 |
+
"grad_norm": 0.2392578125,
|
| 582 |
+
"learning_rate": 0.0003,
|
| 583 |
+
"loss": 0.24779348373413085,
|
| 584 |
+
"step": 1480
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.10111223458038422,
|
| 588 |
+
"grad_norm": 0.236328125,
|
| 589 |
+
"learning_rate": 0.0003,
|
| 590 |
+
"loss": 0.2468804121017456,
|
| 591 |
+
"step": 1500
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.10246039770812268,
|
| 595 |
+
"grad_norm": 0.23828125,
|
| 596 |
+
"learning_rate": 0.0003,
|
| 597 |
+
"loss": 0.2450350522994995,
|
| 598 |
+
"step": 1520
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.10380856083586114,
|
| 602 |
+
"grad_norm": 0.2421875,
|
| 603 |
+
"learning_rate": 0.0003,
|
| 604 |
+
"loss": 0.2447040557861328,
|
| 605 |
+
"step": 1540
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.1051567239635996,
|
| 609 |
+
"grad_norm": 0.240234375,
|
| 610 |
+
"learning_rate": 0.0003,
|
| 611 |
+
"loss": 0.24405441284179688,
|
| 612 |
+
"step": 1560
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.10650488709133805,
|
| 616 |
+
"grad_norm": 0.23828125,
|
| 617 |
+
"learning_rate": 0.0003,
|
| 618 |
+
"loss": 0.24406831264495848,
|
| 619 |
+
"step": 1580
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.10785305021907651,
|
| 623 |
+
"grad_norm": 1.0078125,
|
| 624 |
+
"learning_rate": 0.0003,
|
| 625 |
+
"loss": 0.25422332286834715,
|
| 626 |
+
"step": 1600
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.10785305021907651,
|
| 630 |
+
"eval_loss": 0.2647191882133484,
|
| 631 |
+
"eval_runtime": 9.4144,
|
| 632 |
+
"eval_samples_per_second": 1011.957,
|
| 633 |
+
"eval_steps_per_second": 15.827,
|
| 634 |
+
"step": 1600
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.10920121334681497,
|
| 638 |
+
"grad_norm": 0.384765625,
|
| 639 |
+
"learning_rate": 0.0003,
|
| 640 |
+
"loss": 0.2498933792114258,
|
| 641 |
+
"step": 1620
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.11054937647455342,
|
| 645 |
+
"grad_norm": 0.2451171875,
|
| 646 |
+
"learning_rate": 0.0003,
|
| 647 |
+
"loss": 0.24230828285217285,
|
| 648 |
+
"step": 1640
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 0.11189753960229187,
|
| 652 |
+
"grad_norm": 0.234375,
|
| 653 |
+
"learning_rate": 0.0003,
|
| 654 |
+
"loss": 0.2383427381515503,
|
| 655 |
+
"step": 1660
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 0.11324570273003033,
|
| 659 |
+
"grad_norm": 0.2294921875,
|
| 660 |
+
"learning_rate": 0.0003,
|
| 661 |
+
"loss": 0.23723020553588867,
|
| 662 |
+
"step": 1680
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 0.11459386585776879,
|
| 666 |
+
"grad_norm": 0.228515625,
|
| 667 |
+
"learning_rate": 0.0003,
|
| 668 |
+
"loss": 0.23743033409118652,
|
| 669 |
+
"step": 1700
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.11594202898550725,
|
| 673 |
+
"grad_norm": 0.2255859375,
|
| 674 |
+
"learning_rate": 0.0003,
|
| 675 |
+
"loss": 0.23500254154205322,
|
| 676 |
+
"step": 1720
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.1172901921132457,
|
| 680 |
+
"grad_norm": 0.234375,
|
| 681 |
+
"learning_rate": 0.0003,
|
| 682 |
+
"loss": 0.2353450059890747,
|
| 683 |
+
"step": 1740
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 0.11863835524098416,
|
| 687 |
+
"grad_norm": 0.232421875,
|
| 688 |
+
"learning_rate": 0.0003,
|
| 689 |
+
"loss": 0.23368706703186035,
|
| 690 |
+
"step": 1760
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 0.11998651836872262,
|
| 694 |
+
"grad_norm": 0.23046875,
|
| 695 |
+
"learning_rate": 0.0003,
|
| 696 |
+
"loss": 0.2346323013305664,
|
| 697 |
+
"step": 1780
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 0.12133468149646107,
|
| 701 |
+
"grad_norm": 0.224609375,
|
| 702 |
+
"learning_rate": 0.0003,
|
| 703 |
+
"loss": 0.23471813201904296,
|
| 704 |
+
"step": 1800
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 0.12133468149646107,
|
| 708 |
+
"eval_loss": 0.23367689549922943,
|
| 709 |
+
"eval_runtime": 9.3393,
|
| 710 |
+
"eval_samples_per_second": 1020.093,
|
| 711 |
+
"eval_steps_per_second": 15.954,
|
| 712 |
+
"step": 1800
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 0.12268284462419952,
|
| 716 |
+
"grad_norm": 0.2333984375,
|
| 717 |
+
"learning_rate": 0.0003,
|
| 718 |
+
"loss": 0.23411636352539061,
|
| 719 |
+
"step": 1820
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 0.12403100775193798,
|
| 723 |
+
"grad_norm": 0.2265625,
|
| 724 |
+
"learning_rate": 0.0003,
|
| 725 |
+
"loss": 0.22900819778442383,
|
| 726 |
+
"step": 1840
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 0.12537917087967643,
|
| 730 |
+
"grad_norm": 0.2333984375,
|
| 731 |
+
"learning_rate": 0.0003,
|
| 732 |
+
"loss": 0.22926979064941405,
|
| 733 |
+
"step": 1860
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 0.1267273340074149,
|
| 737 |
+
"grad_norm": 0.2275390625,
|
| 738 |
+
"learning_rate": 0.0003,
|
| 739 |
+
"loss": 0.2291290283203125,
|
| 740 |
+
"step": 1880
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 0.12807549713515334,
|
| 744 |
+
"grad_norm": 0.2314453125,
|
| 745 |
+
"learning_rate": 0.0003,
|
| 746 |
+
"loss": 0.22867155075073242,
|
| 747 |
+
"step": 1900
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 0.12942366026289182,
|
| 751 |
+
"grad_norm": 0.2265625,
|
| 752 |
+
"learning_rate": 0.0003,
|
| 753 |
+
"loss": 0.22638976573944092,
|
| 754 |
+
"step": 1920
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 0.13077182339063026,
|
| 758 |
+
"grad_norm": 0.2294921875,
|
| 759 |
+
"learning_rate": 0.0003,
|
| 760 |
+
"loss": 0.22726998329162598,
|
| 761 |
+
"step": 1940
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 0.13211998651836873,
|
| 765 |
+
"grad_norm": 0.228515625,
|
| 766 |
+
"learning_rate": 0.0003,
|
| 767 |
+
"loss": 0.22477636337280274,
|
| 768 |
+
"step": 1960
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 0.13346814964610718,
|
| 772 |
+
"grad_norm": 0.2265625,
|
| 773 |
+
"learning_rate": 0.0003,
|
| 774 |
+
"loss": 0.2259512186050415,
|
| 775 |
+
"step": 1980
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 0.13481631277384565,
|
| 779 |
+
"grad_norm": 0.2236328125,
|
| 780 |
+
"learning_rate": 0.0003,
|
| 781 |
+
"loss": 0.22533745765686036,
|
| 782 |
+
"step": 2000
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 0.13481631277384565,
|
| 786 |
+
"eval_loss": 0.2269468903541565,
|
| 787 |
+
"eval_runtime": 9.5112,
|
| 788 |
+
"eval_samples_per_second": 1001.659,
|
| 789 |
+
"eval_steps_per_second": 15.666,
|
| 790 |
+
"step": 2000
|
| 791 |
+
}
|
| 792 |
+
],
|
| 793 |
+
"logging_steps": 20,
|
| 794 |
+
"max_steps": 2000,
|
| 795 |
+
"num_input_tokens_seen": 0,
|
| 796 |
+
"num_train_epochs": 1,
|
| 797 |
+
"save_steps": 500,
|
| 798 |
+
"stateful_callbacks": {
|
| 799 |
+
"TrainerControl": {
|
| 800 |
+
"args": {
|
| 801 |
+
"should_epoch_stop": false,
|
| 802 |
+
"should_evaluate": false,
|
| 803 |
+
"should_log": false,
|
| 804 |
+
"should_save": true,
|
| 805 |
+
"should_training_stop": true
|
| 806 |
+
},
|
| 807 |
+
"attributes": {}
|
| 808 |
+
}
|
| 809 |
+
},
|
| 810 |
+
"total_flos": 1355078959104000.0,
|
| 811 |
+
"train_batch_size": 64,
|
| 812 |
+
"trial_name": null,
|
| 813 |
+
"trial_params": null
|
| 814 |
+
}
|
out/glu-relu-21L_run/checkpoint-2000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1f2f1348ca728c5b4f20019812b43d48f864ed7914b38a57a4c9db7bf05ccfd7
|
| 3 |
+
size 4920
|
out/glu-relu-21L_run/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 7961528
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:49014ed1c74d9604971be6d4a4d168cf9abc3c482421411cf1a7bc19b02ca114
|
| 3 |
size 7961528
|
out/glu-relu-21L_run/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/glu-relu-21L_run/tokenizer_config.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": "<|endoftext|>",
|
| 5 |
+
"eos_token": "<|endoftext|>",
|
| 6 |
+
"errors": "replace",
|
| 7 |
+
"is_local": false,
|
| 8 |
+
"local_files_only": false,
|
| 9 |
+
"model_max_length": 1024,
|
| 10 |
+
"pad_token": "<|endoftext|>",
|
| 11 |
+
"tokenizer_class": "GPT2Tokenizer",
|
| 12 |
+
"unk_token": "<|endoftext|>"
|
| 13 |
+
}
|
out/glu-relu-21L_run/training_log.jsonl
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|