Auto upload 2026-08-08T21:16:26.795996 (part 2)
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- out/mlp-sigmoid-46L_run/checkpoint-400/optimizer.pt +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-400/rng_state.pth +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-400/scheduler.pt +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-400/tokenizer.json +0 -0
- out/mlp-sigmoid-46L_run/checkpoint-400/tokenizer_config.json +13 -0
- out/mlp-sigmoid-46L_run/checkpoint-400/trainer_state.json +238 -0
- out/mlp-sigmoid-46L_run/checkpoint-400/training_args.bin +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-500/config.json +35 -0
- out/mlp-sigmoid-46L_run/checkpoint-500/model.safetensors +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-500/optimizer.pt +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-500/rng_state.pth +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-500/scheduler.pt +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-500/tokenizer.json +0 -0
- out/mlp-sigmoid-46L_run/checkpoint-500/tokenizer_config.json +13 -0
- out/mlp-sigmoid-46L_run/checkpoint-500/trainer_state.json +289 -0
- out/mlp-sigmoid-46L_run/checkpoint-500/training_args.bin +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-600/config.json +35 -0
- out/mlp-sigmoid-46L_run/checkpoint-600/model.safetensors +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-600/optimizer.pt +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-600/rng_state.pth +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-600/scheduler.pt +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-600/tokenizer.json +0 -0
- out/mlp-sigmoid-46L_run/checkpoint-600/tokenizer_config.json +13 -0
- out/mlp-sigmoid-46L_run/checkpoint-600/trainer_state.json +340 -0
- out/mlp-sigmoid-46L_run/checkpoint-600/training_args.bin +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-700/config.json +35 -0
- out/mlp-sigmoid-46L_run/checkpoint-700/model.safetensors +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-700/optimizer.pt +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-700/rng_state.pth +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-700/scheduler.pt +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-700/tokenizer.json +0 -0
- out/mlp-sigmoid-46L_run/checkpoint-700/tokenizer_config.json +13 -0
- out/mlp-sigmoid-46L_run/checkpoint-700/trainer_state.json +391 -0
- out/mlp-sigmoid-46L_run/checkpoint-700/training_args.bin +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-800/config.json +35 -0
- out/mlp-sigmoid-46L_run/checkpoint-800/model.safetensors +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-800/optimizer.pt +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-800/rng_state.pth +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-800/scheduler.pt +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-800/tokenizer.json +0 -0
- out/mlp-sigmoid-46L_run/checkpoint-800/tokenizer_config.json +13 -0
- out/mlp-sigmoid-46L_run/checkpoint-800/trainer_state.json +442 -0
- out/mlp-sigmoid-46L_run/checkpoint-800/training_args.bin +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-900/config.json +35 -0
- out/mlp-sigmoid-46L_run/checkpoint-900/model.safetensors +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-900/optimizer.pt +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-900/rng_state.pth +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-900/scheduler.pt +3 -0
- out/mlp-sigmoid-46L_run/checkpoint-900/tokenizer.json +0 -0
- out/mlp-sigmoid-46L_run/checkpoint-900/tokenizer_config.json +13 -0
out/mlp-sigmoid-46L_run/checkpoint-400/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8b1445029ef219af5bc17363d99b153600f9df0dc6e06700c1285a259ceb1493
|
| 3 |
+
size 32605416
|
out/mlp-sigmoid-46L_run/checkpoint-400/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f9a6944c405a002fce05f295d08ea6650e2e2ad6dbf5d6da1e9053f7bf7f5827
|
| 3 |
+
size 14244
|
out/mlp-sigmoid-46L_run/checkpoint-400/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:51769a11c87b2b665cbe64c58b934afdb1fa1998bffb4addcc2f852b172d6681
|
| 3 |
+
size 1064
|
out/mlp-sigmoid-46L_run/checkpoint-400/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/mlp-sigmoid-46L_run/checkpoint-400/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/mlp-sigmoid-46L_run/checkpoint-400/trainer_state.json
ADDED
|
@@ -0,0 +1,238 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.215633423180593,
|
| 6 |
+
"eval_steps": 50,
|
| 7 |
+
"global_step": 400,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01078167115902965,
|
| 14 |
+
"grad_norm": 0.953125,
|
| 15 |
+
"learning_rate": 0.001,
|
| 16 |
+
"loss": 7.132730865478516,
|
| 17 |
+
"step": 20
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0215633423180593,
|
| 21 |
+
"grad_norm": 0.30078125,
|
| 22 |
+
"learning_rate": 0.001,
|
| 23 |
+
"loss": 6.084678268432617,
|
| 24 |
+
"step": 40
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.026954177897574125,
|
| 28 |
+
"eval_loss": 5.9758501052856445,
|
| 29 |
+
"eval_runtime": 11.1232,
|
| 30 |
+
"eval_samples_per_second": 856.498,
|
| 31 |
+
"eval_steps_per_second": 1.708,
|
| 32 |
+
"step": 50
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.03234501347708895,
|
| 36 |
+
"grad_norm": 0.4140625,
|
| 37 |
+
"learning_rate": 0.001,
|
| 38 |
+
"loss": 5.959280776977539,
|
| 39 |
+
"step": 60
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.0431266846361186,
|
| 43 |
+
"grad_norm": 0.2333984375,
|
| 44 |
+
"learning_rate": 0.001,
|
| 45 |
+
"loss": 5.698427963256836,
|
| 46 |
+
"step": 80
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.05390835579514825,
|
| 50 |
+
"grad_norm": 0.26953125,
|
| 51 |
+
"learning_rate": 0.001,
|
| 52 |
+
"loss": 5.483244323730469,
|
| 53 |
+
"step": 100
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.05390835579514825,
|
| 57 |
+
"eval_loss": 5.370211601257324,
|
| 58 |
+
"eval_runtime": 11.0192,
|
| 59 |
+
"eval_samples_per_second": 864.582,
|
| 60 |
+
"eval_steps_per_second": 1.724,
|
| 61 |
+
"step": 100
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.0646900269541779,
|
| 65 |
+
"grad_norm": 0.33203125,
|
| 66 |
+
"learning_rate": 0.001,
|
| 67 |
+
"loss": 5.284563827514648,
|
| 68 |
+
"step": 120
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.07547169811320754,
|
| 72 |
+
"grad_norm": 0.5859375,
|
| 73 |
+
"learning_rate": 0.001,
|
| 74 |
+
"loss": 5.074047470092774,
|
| 75 |
+
"step": 140
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.08086253369272237,
|
| 79 |
+
"eval_loss": 4.84706974029541,
|
| 80 |
+
"eval_runtime": 11.0423,
|
| 81 |
+
"eval_samples_per_second": 862.77,
|
| 82 |
+
"eval_steps_per_second": 1.721,
|
| 83 |
+
"step": 150
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.0862533692722372,
|
| 87 |
+
"grad_norm": 0.578125,
|
| 88 |
+
"learning_rate": 0.001,
|
| 89 |
+
"loss": 4.856560516357422,
|
| 90 |
+
"step": 160
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.09703504043126684,
|
| 94 |
+
"grad_norm": 0.76953125,
|
| 95 |
+
"learning_rate": 0.001,
|
| 96 |
+
"loss": 4.646049499511719,
|
| 97 |
+
"step": 180
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.1078167115902965,
|
| 101 |
+
"grad_norm": 0.32421875,
|
| 102 |
+
"learning_rate": 0.001,
|
| 103 |
+
"loss": 4.480390548706055,
|
| 104 |
+
"step": 200
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.1078167115902965,
|
| 108 |
+
"eval_loss": 4.395232677459717,
|
| 109 |
+
"eval_runtime": 11.0603,
|
| 110 |
+
"eval_samples_per_second": 861.369,
|
| 111 |
+
"eval_steps_per_second": 1.718,
|
| 112 |
+
"step": 200
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 0.11859838274932614,
|
| 116 |
+
"grad_norm": 0.57421875,
|
| 117 |
+
"learning_rate": 0.001,
|
| 118 |
+
"loss": 4.35283317565918,
|
| 119 |
+
"step": 220
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 0.1293800539083558,
|
| 123 |
+
"grad_norm": 0.64453125,
|
| 124 |
+
"learning_rate": 0.001,
|
| 125 |
+
"loss": 4.239774703979492,
|
| 126 |
+
"step": 240
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 0.1347708894878706,
|
| 130 |
+
"eval_loss": 4.130031108856201,
|
| 131 |
+
"eval_runtime": 11.5228,
|
| 132 |
+
"eval_samples_per_second": 826.793,
|
| 133 |
+
"eval_steps_per_second": 1.649,
|
| 134 |
+
"step": 250
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.14016172506738545,
|
| 138 |
+
"grad_norm": 1.3046875,
|
| 139 |
+
"learning_rate": 0.001,
|
| 140 |
+
"loss": 4.1399494171142575,
|
| 141 |
+
"step": 260
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.1509433962264151,
|
| 145 |
+
"grad_norm": 0.68359375,
|
| 146 |
+
"learning_rate": 0.001,
|
| 147 |
+
"loss": 4.071997833251953,
|
| 148 |
+
"step": 280
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 0.16172506738544473,
|
| 152 |
+
"grad_norm": 0.6484375,
|
| 153 |
+
"learning_rate": 0.001,
|
| 154 |
+
"loss": 4.0012977600097654,
|
| 155 |
+
"step": 300
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 0.16172506738544473,
|
| 159 |
+
"eval_loss": 3.970766067504883,
|
| 160 |
+
"eval_runtime": 11.2668,
|
| 161 |
+
"eval_samples_per_second": 845.58,
|
| 162 |
+
"eval_steps_per_second": 1.686,
|
| 163 |
+
"step": 300
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.1725067385444744,
|
| 167 |
+
"grad_norm": 0.494140625,
|
| 168 |
+
"learning_rate": 0.001,
|
| 169 |
+
"loss": 3.9426082611083983,
|
| 170 |
+
"step": 320
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.18328840970350405,
|
| 174 |
+
"grad_norm": 0.3359375,
|
| 175 |
+
"learning_rate": 0.001,
|
| 176 |
+
"loss": 3.884596252441406,
|
| 177 |
+
"step": 340
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.18867924528301888,
|
| 181 |
+
"eval_loss": 3.869180202484131,
|
| 182 |
+
"eval_runtime": 11.0104,
|
| 183 |
+
"eval_samples_per_second": 865.274,
|
| 184 |
+
"eval_steps_per_second": 1.726,
|
| 185 |
+
"step": 350
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.1940700808625337,
|
| 189 |
+
"grad_norm": 0.59765625,
|
| 190 |
+
"learning_rate": 0.001,
|
| 191 |
+
"loss": 3.846722412109375,
|
| 192 |
+
"step": 360
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.20485175202156333,
|
| 196 |
+
"grad_norm": 0.3203125,
|
| 197 |
+
"learning_rate": 0.001,
|
| 198 |
+
"loss": 3.7851001739501955,
|
| 199 |
+
"step": 380
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.215633423180593,
|
| 203 |
+
"grad_norm": 0.423828125,
|
| 204 |
+
"learning_rate": 0.001,
|
| 205 |
+
"loss": 3.758561706542969,
|
| 206 |
+
"step": 400
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.215633423180593,
|
| 210 |
+
"eval_loss": 3.7298622131347656,
|
| 211 |
+
"eval_runtime": 11.1,
|
| 212 |
+
"eval_samples_per_second": 858.29,
|
| 213 |
+
"eval_steps_per_second": 1.712,
|
| 214 |
+
"step": 400
|
| 215 |
+
}
|
| 216 |
+
],
|
| 217 |
+
"logging_steps": 20,
|
| 218 |
+
"max_steps": 1000,
|
| 219 |
+
"num_input_tokens_seen": 0,
|
| 220 |
+
"num_train_epochs": 1,
|
| 221 |
+
"save_steps": 100,
|
| 222 |
+
"stateful_callbacks": {
|
| 223 |
+
"TrainerControl": {
|
| 224 |
+
"args": {
|
| 225 |
+
"should_epoch_stop": false,
|
| 226 |
+
"should_evaluate": false,
|
| 227 |
+
"should_log": false,
|
| 228 |
+
"should_save": true,
|
| 229 |
+
"should_training_stop": false
|
| 230 |
+
},
|
| 231 |
+
"attributes": {}
|
| 232 |
+
}
|
| 233 |
+
},
|
| 234 |
+
"total_flos": 4749133244006400.0,
|
| 235 |
+
"train_batch_size": 512,
|
| 236 |
+
"trial_name": null,
|
| 237 |
+
"trial_params": null
|
| 238 |
+
}
|
out/mlp-sigmoid-46L_run/checkpoint-400/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f497915d25e5adfc7823f2b88d2f779dbbdb1cedd0353e013bfd915138617020
|
| 3 |
+
size 4920
|
out/mlp-sigmoid-46L_run/checkpoint-500/config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation": "sigmoid",
|
| 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": "mlp",
|
| 19 |
+
"model_type": "tiny_llama",
|
| 20 |
+
"num_attention_heads": 4,
|
| 21 |
+
"num_hidden_layers": 46,
|
| 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/mlp-sigmoid-46L_run/checkpoint-500/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e34574332a97a093e666133bc052f18ea6521d1391a8ae7961a89ed251b6f444
|
| 3 |
+
size 16186168
|
out/mlp-sigmoid-46L_run/checkpoint-500/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:19b64bac0b981846aaa9bb3dd97db491c7fa5115262022fd77ed043e750438a6
|
| 3 |
+
size 32605416
|
out/mlp-sigmoid-46L_run/checkpoint-500/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/mlp-sigmoid-46L_run/checkpoint-500/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:60a0752f1683b6bad1b285850c7a77f8c14fbcad243554c3a99f6e443a78365e
|
| 3 |
+
size 1064
|
out/mlp-sigmoid-46L_run/checkpoint-500/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/mlp-sigmoid-46L_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/mlp-sigmoid-46L_run/checkpoint-500/trainer_state.json
ADDED
|
@@ -0,0 +1,289 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.2695417789757412,
|
| 6 |
+
"eval_steps": 50,
|
| 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": 0.01078167115902965,
|
| 14 |
+
"grad_norm": 0.953125,
|
| 15 |
+
"learning_rate": 0.001,
|
| 16 |
+
"loss": 7.132730865478516,
|
| 17 |
+
"step": 20
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0215633423180593,
|
| 21 |
+
"grad_norm": 0.30078125,
|
| 22 |
+
"learning_rate": 0.001,
|
| 23 |
+
"loss": 6.084678268432617,
|
| 24 |
+
"step": 40
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.026954177897574125,
|
| 28 |
+
"eval_loss": 5.9758501052856445,
|
| 29 |
+
"eval_runtime": 11.1232,
|
| 30 |
+
"eval_samples_per_second": 856.498,
|
| 31 |
+
"eval_steps_per_second": 1.708,
|
| 32 |
+
"step": 50
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.03234501347708895,
|
| 36 |
+
"grad_norm": 0.4140625,
|
| 37 |
+
"learning_rate": 0.001,
|
| 38 |
+
"loss": 5.959280776977539,
|
| 39 |
+
"step": 60
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.0431266846361186,
|
| 43 |
+
"grad_norm": 0.2333984375,
|
| 44 |
+
"learning_rate": 0.001,
|
| 45 |
+
"loss": 5.698427963256836,
|
| 46 |
+
"step": 80
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.05390835579514825,
|
| 50 |
+
"grad_norm": 0.26953125,
|
| 51 |
+
"learning_rate": 0.001,
|
| 52 |
+
"loss": 5.483244323730469,
|
| 53 |
+
"step": 100
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.05390835579514825,
|
| 57 |
+
"eval_loss": 5.370211601257324,
|
| 58 |
+
"eval_runtime": 11.0192,
|
| 59 |
+
"eval_samples_per_second": 864.582,
|
| 60 |
+
"eval_steps_per_second": 1.724,
|
| 61 |
+
"step": 100
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.0646900269541779,
|
| 65 |
+
"grad_norm": 0.33203125,
|
| 66 |
+
"learning_rate": 0.001,
|
| 67 |
+
"loss": 5.284563827514648,
|
| 68 |
+
"step": 120
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.07547169811320754,
|
| 72 |
+
"grad_norm": 0.5859375,
|
| 73 |
+
"learning_rate": 0.001,
|
| 74 |
+
"loss": 5.074047470092774,
|
| 75 |
+
"step": 140
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.08086253369272237,
|
| 79 |
+
"eval_loss": 4.84706974029541,
|
| 80 |
+
"eval_runtime": 11.0423,
|
| 81 |
+
"eval_samples_per_second": 862.77,
|
| 82 |
+
"eval_steps_per_second": 1.721,
|
| 83 |
+
"step": 150
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.0862533692722372,
|
| 87 |
+
"grad_norm": 0.578125,
|
| 88 |
+
"learning_rate": 0.001,
|
| 89 |
+
"loss": 4.856560516357422,
|
| 90 |
+
"step": 160
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.09703504043126684,
|
| 94 |
+
"grad_norm": 0.76953125,
|
| 95 |
+
"learning_rate": 0.001,
|
| 96 |
+
"loss": 4.646049499511719,
|
| 97 |
+
"step": 180
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.1078167115902965,
|
| 101 |
+
"grad_norm": 0.32421875,
|
| 102 |
+
"learning_rate": 0.001,
|
| 103 |
+
"loss": 4.480390548706055,
|
| 104 |
+
"step": 200
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.1078167115902965,
|
| 108 |
+
"eval_loss": 4.395232677459717,
|
| 109 |
+
"eval_runtime": 11.0603,
|
| 110 |
+
"eval_samples_per_second": 861.369,
|
| 111 |
+
"eval_steps_per_second": 1.718,
|
| 112 |
+
"step": 200
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 0.11859838274932614,
|
| 116 |
+
"grad_norm": 0.57421875,
|
| 117 |
+
"learning_rate": 0.001,
|
| 118 |
+
"loss": 4.35283317565918,
|
| 119 |
+
"step": 220
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 0.1293800539083558,
|
| 123 |
+
"grad_norm": 0.64453125,
|
| 124 |
+
"learning_rate": 0.001,
|
| 125 |
+
"loss": 4.239774703979492,
|
| 126 |
+
"step": 240
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 0.1347708894878706,
|
| 130 |
+
"eval_loss": 4.130031108856201,
|
| 131 |
+
"eval_runtime": 11.5228,
|
| 132 |
+
"eval_samples_per_second": 826.793,
|
| 133 |
+
"eval_steps_per_second": 1.649,
|
| 134 |
+
"step": 250
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.14016172506738545,
|
| 138 |
+
"grad_norm": 1.3046875,
|
| 139 |
+
"learning_rate": 0.001,
|
| 140 |
+
"loss": 4.1399494171142575,
|
| 141 |
+
"step": 260
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.1509433962264151,
|
| 145 |
+
"grad_norm": 0.68359375,
|
| 146 |
+
"learning_rate": 0.001,
|
| 147 |
+
"loss": 4.071997833251953,
|
| 148 |
+
"step": 280
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 0.16172506738544473,
|
| 152 |
+
"grad_norm": 0.6484375,
|
| 153 |
+
"learning_rate": 0.001,
|
| 154 |
+
"loss": 4.0012977600097654,
|
| 155 |
+
"step": 300
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 0.16172506738544473,
|
| 159 |
+
"eval_loss": 3.970766067504883,
|
| 160 |
+
"eval_runtime": 11.2668,
|
| 161 |
+
"eval_samples_per_second": 845.58,
|
| 162 |
+
"eval_steps_per_second": 1.686,
|
| 163 |
+
"step": 300
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.1725067385444744,
|
| 167 |
+
"grad_norm": 0.494140625,
|
| 168 |
+
"learning_rate": 0.001,
|
| 169 |
+
"loss": 3.9426082611083983,
|
| 170 |
+
"step": 320
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.18328840970350405,
|
| 174 |
+
"grad_norm": 0.3359375,
|
| 175 |
+
"learning_rate": 0.001,
|
| 176 |
+
"loss": 3.884596252441406,
|
| 177 |
+
"step": 340
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.18867924528301888,
|
| 181 |
+
"eval_loss": 3.869180202484131,
|
| 182 |
+
"eval_runtime": 11.0104,
|
| 183 |
+
"eval_samples_per_second": 865.274,
|
| 184 |
+
"eval_steps_per_second": 1.726,
|
| 185 |
+
"step": 350
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.1940700808625337,
|
| 189 |
+
"grad_norm": 0.59765625,
|
| 190 |
+
"learning_rate": 0.001,
|
| 191 |
+
"loss": 3.846722412109375,
|
| 192 |
+
"step": 360
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.20485175202156333,
|
| 196 |
+
"grad_norm": 0.3203125,
|
| 197 |
+
"learning_rate": 0.001,
|
| 198 |
+
"loss": 3.7851001739501955,
|
| 199 |
+
"step": 380
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.215633423180593,
|
| 203 |
+
"grad_norm": 0.423828125,
|
| 204 |
+
"learning_rate": 0.001,
|
| 205 |
+
"loss": 3.758561706542969,
|
| 206 |
+
"step": 400
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.215633423180593,
|
| 210 |
+
"eval_loss": 3.7298622131347656,
|
| 211 |
+
"eval_runtime": 11.1,
|
| 212 |
+
"eval_samples_per_second": 858.29,
|
| 213 |
+
"eval_steps_per_second": 1.712,
|
| 214 |
+
"step": 400
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 0.22641509433962265,
|
| 218 |
+
"grad_norm": 0.330078125,
|
| 219 |
+
"learning_rate": 0.001,
|
| 220 |
+
"loss": 3.7036609649658203,
|
| 221 |
+
"step": 420
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.2371967654986523,
|
| 225 |
+
"grad_norm": 0.69140625,
|
| 226 |
+
"learning_rate": 0.001,
|
| 227 |
+
"loss": 3.6900096893310548,
|
| 228 |
+
"step": 440
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 0.24258760107816713,
|
| 232 |
+
"eval_loss": 3.649232864379883,
|
| 233 |
+
"eval_runtime": 11.3447,
|
| 234 |
+
"eval_samples_per_second": 839.777,
|
| 235 |
+
"eval_steps_per_second": 1.675,
|
| 236 |
+
"step": 450
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.24797843665768193,
|
| 240 |
+
"grad_norm": 0.42578125,
|
| 241 |
+
"learning_rate": 0.001,
|
| 242 |
+
"loss": 3.65367431640625,
|
| 243 |
+
"step": 460
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.2587601078167116,
|
| 247 |
+
"grad_norm": 0.70703125,
|
| 248 |
+
"learning_rate": 0.001,
|
| 249 |
+
"loss": 3.617287826538086,
|
| 250 |
+
"step": 480
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.2695417789757412,
|
| 254 |
+
"grad_norm": 0.68359375,
|
| 255 |
+
"learning_rate": 0.001,
|
| 256 |
+
"loss": 3.580358123779297,
|
| 257 |
+
"step": 500
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.2695417789757412,
|
| 261 |
+
"eval_loss": 3.590764284133911,
|
| 262 |
+
"eval_runtime": 11.1385,
|
| 263 |
+
"eval_samples_per_second": 855.32,
|
| 264 |
+
"eval_steps_per_second": 1.706,
|
| 265 |
+
"step": 500
|
| 266 |
+
}
|
| 267 |
+
],
|
| 268 |
+
"logging_steps": 20,
|
| 269 |
+
"max_steps": 1000,
|
| 270 |
+
"num_input_tokens_seen": 0,
|
| 271 |
+
"num_train_epochs": 1,
|
| 272 |
+
"save_steps": 100,
|
| 273 |
+
"stateful_callbacks": {
|
| 274 |
+
"TrainerControl": {
|
| 275 |
+
"args": {
|
| 276 |
+
"should_epoch_stop": false,
|
| 277 |
+
"should_evaluate": false,
|
| 278 |
+
"should_log": false,
|
| 279 |
+
"should_save": true,
|
| 280 |
+
"should_training_stop": false
|
| 281 |
+
},
|
| 282 |
+
"attributes": {}
|
| 283 |
+
}
|
| 284 |
+
},
|
| 285 |
+
"total_flos": 5936416555008000.0,
|
| 286 |
+
"train_batch_size": 512,
|
| 287 |
+
"trial_name": null,
|
| 288 |
+
"trial_params": null
|
| 289 |
+
}
|
out/mlp-sigmoid-46L_run/checkpoint-500/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f497915d25e5adfc7823f2b88d2f779dbbdb1cedd0353e013bfd915138617020
|
| 3 |
+
size 4920
|
out/mlp-sigmoid-46L_run/checkpoint-600/config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation": "sigmoid",
|
| 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": "mlp",
|
| 19 |
+
"model_type": "tiny_llama",
|
| 20 |
+
"num_attention_heads": 4,
|
| 21 |
+
"num_hidden_layers": 46,
|
| 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/mlp-sigmoid-46L_run/checkpoint-600/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9f9585bff5b3044485209e27d07689f2a38fff9f348f79ab82c078cefd86653b
|
| 3 |
+
size 16186168
|
out/mlp-sigmoid-46L_run/checkpoint-600/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d0f53815c9bc3f4a572864b426f8912c1f7877940347b5e99099cf6057a4024d
|
| 3 |
+
size 32605416
|
out/mlp-sigmoid-46L_run/checkpoint-600/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b0b8a6e3878f9402461193b241d8f1ee8546c6ad03f43e5b9dab8f4fc8c4d065
|
| 3 |
+
size 14244
|
out/mlp-sigmoid-46L_run/checkpoint-600/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c5502244c7ba8b61ab0a3b01cdc7e5e2629060a0565d6b782445d6ba31022cee
|
| 3 |
+
size 1064
|
out/mlp-sigmoid-46L_run/checkpoint-600/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/mlp-sigmoid-46L_run/checkpoint-600/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/mlp-sigmoid-46L_run/checkpoint-600/trainer_state.json
ADDED
|
@@ -0,0 +1,340 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.32345013477088946,
|
| 6 |
+
"eval_steps": 50,
|
| 7 |
+
"global_step": 600,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01078167115902965,
|
| 14 |
+
"grad_norm": 0.953125,
|
| 15 |
+
"learning_rate": 0.001,
|
| 16 |
+
"loss": 7.132730865478516,
|
| 17 |
+
"step": 20
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0215633423180593,
|
| 21 |
+
"grad_norm": 0.30078125,
|
| 22 |
+
"learning_rate": 0.001,
|
| 23 |
+
"loss": 6.084678268432617,
|
| 24 |
+
"step": 40
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.026954177897574125,
|
| 28 |
+
"eval_loss": 5.9758501052856445,
|
| 29 |
+
"eval_runtime": 11.1232,
|
| 30 |
+
"eval_samples_per_second": 856.498,
|
| 31 |
+
"eval_steps_per_second": 1.708,
|
| 32 |
+
"step": 50
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.03234501347708895,
|
| 36 |
+
"grad_norm": 0.4140625,
|
| 37 |
+
"learning_rate": 0.001,
|
| 38 |
+
"loss": 5.959280776977539,
|
| 39 |
+
"step": 60
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.0431266846361186,
|
| 43 |
+
"grad_norm": 0.2333984375,
|
| 44 |
+
"learning_rate": 0.001,
|
| 45 |
+
"loss": 5.698427963256836,
|
| 46 |
+
"step": 80
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.05390835579514825,
|
| 50 |
+
"grad_norm": 0.26953125,
|
| 51 |
+
"learning_rate": 0.001,
|
| 52 |
+
"loss": 5.483244323730469,
|
| 53 |
+
"step": 100
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.05390835579514825,
|
| 57 |
+
"eval_loss": 5.370211601257324,
|
| 58 |
+
"eval_runtime": 11.0192,
|
| 59 |
+
"eval_samples_per_second": 864.582,
|
| 60 |
+
"eval_steps_per_second": 1.724,
|
| 61 |
+
"step": 100
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.0646900269541779,
|
| 65 |
+
"grad_norm": 0.33203125,
|
| 66 |
+
"learning_rate": 0.001,
|
| 67 |
+
"loss": 5.284563827514648,
|
| 68 |
+
"step": 120
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.07547169811320754,
|
| 72 |
+
"grad_norm": 0.5859375,
|
| 73 |
+
"learning_rate": 0.001,
|
| 74 |
+
"loss": 5.074047470092774,
|
| 75 |
+
"step": 140
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.08086253369272237,
|
| 79 |
+
"eval_loss": 4.84706974029541,
|
| 80 |
+
"eval_runtime": 11.0423,
|
| 81 |
+
"eval_samples_per_second": 862.77,
|
| 82 |
+
"eval_steps_per_second": 1.721,
|
| 83 |
+
"step": 150
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.0862533692722372,
|
| 87 |
+
"grad_norm": 0.578125,
|
| 88 |
+
"learning_rate": 0.001,
|
| 89 |
+
"loss": 4.856560516357422,
|
| 90 |
+
"step": 160
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.09703504043126684,
|
| 94 |
+
"grad_norm": 0.76953125,
|
| 95 |
+
"learning_rate": 0.001,
|
| 96 |
+
"loss": 4.646049499511719,
|
| 97 |
+
"step": 180
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.1078167115902965,
|
| 101 |
+
"grad_norm": 0.32421875,
|
| 102 |
+
"learning_rate": 0.001,
|
| 103 |
+
"loss": 4.480390548706055,
|
| 104 |
+
"step": 200
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.1078167115902965,
|
| 108 |
+
"eval_loss": 4.395232677459717,
|
| 109 |
+
"eval_runtime": 11.0603,
|
| 110 |
+
"eval_samples_per_second": 861.369,
|
| 111 |
+
"eval_steps_per_second": 1.718,
|
| 112 |
+
"step": 200
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 0.11859838274932614,
|
| 116 |
+
"grad_norm": 0.57421875,
|
| 117 |
+
"learning_rate": 0.001,
|
| 118 |
+
"loss": 4.35283317565918,
|
| 119 |
+
"step": 220
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 0.1293800539083558,
|
| 123 |
+
"grad_norm": 0.64453125,
|
| 124 |
+
"learning_rate": 0.001,
|
| 125 |
+
"loss": 4.239774703979492,
|
| 126 |
+
"step": 240
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 0.1347708894878706,
|
| 130 |
+
"eval_loss": 4.130031108856201,
|
| 131 |
+
"eval_runtime": 11.5228,
|
| 132 |
+
"eval_samples_per_second": 826.793,
|
| 133 |
+
"eval_steps_per_second": 1.649,
|
| 134 |
+
"step": 250
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.14016172506738545,
|
| 138 |
+
"grad_norm": 1.3046875,
|
| 139 |
+
"learning_rate": 0.001,
|
| 140 |
+
"loss": 4.1399494171142575,
|
| 141 |
+
"step": 260
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.1509433962264151,
|
| 145 |
+
"grad_norm": 0.68359375,
|
| 146 |
+
"learning_rate": 0.001,
|
| 147 |
+
"loss": 4.071997833251953,
|
| 148 |
+
"step": 280
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 0.16172506738544473,
|
| 152 |
+
"grad_norm": 0.6484375,
|
| 153 |
+
"learning_rate": 0.001,
|
| 154 |
+
"loss": 4.0012977600097654,
|
| 155 |
+
"step": 300
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 0.16172506738544473,
|
| 159 |
+
"eval_loss": 3.970766067504883,
|
| 160 |
+
"eval_runtime": 11.2668,
|
| 161 |
+
"eval_samples_per_second": 845.58,
|
| 162 |
+
"eval_steps_per_second": 1.686,
|
| 163 |
+
"step": 300
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.1725067385444744,
|
| 167 |
+
"grad_norm": 0.494140625,
|
| 168 |
+
"learning_rate": 0.001,
|
| 169 |
+
"loss": 3.9426082611083983,
|
| 170 |
+
"step": 320
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.18328840970350405,
|
| 174 |
+
"grad_norm": 0.3359375,
|
| 175 |
+
"learning_rate": 0.001,
|
| 176 |
+
"loss": 3.884596252441406,
|
| 177 |
+
"step": 340
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.18867924528301888,
|
| 181 |
+
"eval_loss": 3.869180202484131,
|
| 182 |
+
"eval_runtime": 11.0104,
|
| 183 |
+
"eval_samples_per_second": 865.274,
|
| 184 |
+
"eval_steps_per_second": 1.726,
|
| 185 |
+
"step": 350
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.1940700808625337,
|
| 189 |
+
"grad_norm": 0.59765625,
|
| 190 |
+
"learning_rate": 0.001,
|
| 191 |
+
"loss": 3.846722412109375,
|
| 192 |
+
"step": 360
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.20485175202156333,
|
| 196 |
+
"grad_norm": 0.3203125,
|
| 197 |
+
"learning_rate": 0.001,
|
| 198 |
+
"loss": 3.7851001739501955,
|
| 199 |
+
"step": 380
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.215633423180593,
|
| 203 |
+
"grad_norm": 0.423828125,
|
| 204 |
+
"learning_rate": 0.001,
|
| 205 |
+
"loss": 3.758561706542969,
|
| 206 |
+
"step": 400
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.215633423180593,
|
| 210 |
+
"eval_loss": 3.7298622131347656,
|
| 211 |
+
"eval_runtime": 11.1,
|
| 212 |
+
"eval_samples_per_second": 858.29,
|
| 213 |
+
"eval_steps_per_second": 1.712,
|
| 214 |
+
"step": 400
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 0.22641509433962265,
|
| 218 |
+
"grad_norm": 0.330078125,
|
| 219 |
+
"learning_rate": 0.001,
|
| 220 |
+
"loss": 3.7036609649658203,
|
| 221 |
+
"step": 420
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.2371967654986523,
|
| 225 |
+
"grad_norm": 0.69140625,
|
| 226 |
+
"learning_rate": 0.001,
|
| 227 |
+
"loss": 3.6900096893310548,
|
| 228 |
+
"step": 440
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 0.24258760107816713,
|
| 232 |
+
"eval_loss": 3.649232864379883,
|
| 233 |
+
"eval_runtime": 11.3447,
|
| 234 |
+
"eval_samples_per_second": 839.777,
|
| 235 |
+
"eval_steps_per_second": 1.675,
|
| 236 |
+
"step": 450
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.24797843665768193,
|
| 240 |
+
"grad_norm": 0.42578125,
|
| 241 |
+
"learning_rate": 0.001,
|
| 242 |
+
"loss": 3.65367431640625,
|
| 243 |
+
"step": 460
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.2587601078167116,
|
| 247 |
+
"grad_norm": 0.70703125,
|
| 248 |
+
"learning_rate": 0.001,
|
| 249 |
+
"loss": 3.617287826538086,
|
| 250 |
+
"step": 480
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.2695417789757412,
|
| 254 |
+
"grad_norm": 0.68359375,
|
| 255 |
+
"learning_rate": 0.001,
|
| 256 |
+
"loss": 3.580358123779297,
|
| 257 |
+
"step": 500
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.2695417789757412,
|
| 261 |
+
"eval_loss": 3.590764284133911,
|
| 262 |
+
"eval_runtime": 11.1385,
|
| 263 |
+
"eval_samples_per_second": 855.32,
|
| 264 |
+
"eval_steps_per_second": 1.706,
|
| 265 |
+
"step": 500
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.2803234501347709,
|
| 269 |
+
"grad_norm": 0.373046875,
|
| 270 |
+
"learning_rate": 0.001,
|
| 271 |
+
"loss": 3.557994079589844,
|
| 272 |
+
"step": 520
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.29110512129380056,
|
| 276 |
+
"grad_norm": 0.671875,
|
| 277 |
+
"learning_rate": 0.001,
|
| 278 |
+
"loss": 3.5252918243408202,
|
| 279 |
+
"step": 540
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.29649595687331537,
|
| 283 |
+
"eval_loss": 3.510967969894409,
|
| 284 |
+
"eval_runtime": 11.0988,
|
| 285 |
+
"eval_samples_per_second": 858.384,
|
| 286 |
+
"eval_steps_per_second": 1.712,
|
| 287 |
+
"step": 550
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.3018867924528302,
|
| 291 |
+
"grad_norm": 0.34375,
|
| 292 |
+
"learning_rate": 0.001,
|
| 293 |
+
"loss": 3.5099483489990235,
|
| 294 |
+
"step": 560
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.31266846361185985,
|
| 298 |
+
"grad_norm": 0.77734375,
|
| 299 |
+
"learning_rate": 0.001,
|
| 300 |
+
"loss": 3.471272277832031,
|
| 301 |
+
"step": 580
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.32345013477088946,
|
| 305 |
+
"grad_norm": 0.62109375,
|
| 306 |
+
"learning_rate": 0.001,
|
| 307 |
+
"loss": 3.477389907836914,
|
| 308 |
+
"step": 600
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.32345013477088946,
|
| 312 |
+
"eval_loss": 3.446561098098755,
|
| 313 |
+
"eval_runtime": 11.1795,
|
| 314 |
+
"eval_samples_per_second": 852.182,
|
| 315 |
+
"eval_steps_per_second": 1.7,
|
| 316 |
+
"step": 600
|
| 317 |
+
}
|
| 318 |
+
],
|
| 319 |
+
"logging_steps": 20,
|
| 320 |
+
"max_steps": 1000,
|
| 321 |
+
"num_input_tokens_seen": 0,
|
| 322 |
+
"num_train_epochs": 1,
|
| 323 |
+
"save_steps": 100,
|
| 324 |
+
"stateful_callbacks": {
|
| 325 |
+
"TrainerControl": {
|
| 326 |
+
"args": {
|
| 327 |
+
"should_epoch_stop": false,
|
| 328 |
+
"should_evaluate": false,
|
| 329 |
+
"should_log": false,
|
| 330 |
+
"should_save": true,
|
| 331 |
+
"should_training_stop": false
|
| 332 |
+
},
|
| 333 |
+
"attributes": {}
|
| 334 |
+
}
|
| 335 |
+
},
|
| 336 |
+
"total_flos": 7123699866009600.0,
|
| 337 |
+
"train_batch_size": 512,
|
| 338 |
+
"trial_name": null,
|
| 339 |
+
"trial_params": null
|
| 340 |
+
}
|
out/mlp-sigmoid-46L_run/checkpoint-600/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f497915d25e5adfc7823f2b88d2f779dbbdb1cedd0353e013bfd915138617020
|
| 3 |
+
size 4920
|
out/mlp-sigmoid-46L_run/checkpoint-700/config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation": "sigmoid",
|
| 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": "mlp",
|
| 19 |
+
"model_type": "tiny_llama",
|
| 20 |
+
"num_attention_heads": 4,
|
| 21 |
+
"num_hidden_layers": 46,
|
| 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/mlp-sigmoid-46L_run/checkpoint-700/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:49db0bc4aff1ff4d42d9fe6df78a4228b70620a1287c09cb4f032d953924ed9b
|
| 3 |
+
size 16186168
|
out/mlp-sigmoid-46L_run/checkpoint-700/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:76ad7af55a7a589fed4094fd85676d05bff3893a0128086962008b74a20596d0
|
| 3 |
+
size 32605416
|
out/mlp-sigmoid-46L_run/checkpoint-700/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c07c9483d2aaa0e0aa4859fa05bacc55a60e0f30ff9c95a3b76854e880483a96
|
| 3 |
+
size 14244
|
out/mlp-sigmoid-46L_run/checkpoint-700/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3173cd6908322b02fdacbfeb79bbb6a1a1a40d218262cceb79ee8c11c4fa07d7
|
| 3 |
+
size 1064
|
out/mlp-sigmoid-46L_run/checkpoint-700/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/mlp-sigmoid-46L_run/checkpoint-700/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/mlp-sigmoid-46L_run/checkpoint-700/trainer_state.json
ADDED
|
@@ -0,0 +1,391 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.37735849056603776,
|
| 6 |
+
"eval_steps": 50,
|
| 7 |
+
"global_step": 700,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01078167115902965,
|
| 14 |
+
"grad_norm": 0.953125,
|
| 15 |
+
"learning_rate": 0.001,
|
| 16 |
+
"loss": 7.132730865478516,
|
| 17 |
+
"step": 20
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0215633423180593,
|
| 21 |
+
"grad_norm": 0.30078125,
|
| 22 |
+
"learning_rate": 0.001,
|
| 23 |
+
"loss": 6.084678268432617,
|
| 24 |
+
"step": 40
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.026954177897574125,
|
| 28 |
+
"eval_loss": 5.9758501052856445,
|
| 29 |
+
"eval_runtime": 11.1232,
|
| 30 |
+
"eval_samples_per_second": 856.498,
|
| 31 |
+
"eval_steps_per_second": 1.708,
|
| 32 |
+
"step": 50
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.03234501347708895,
|
| 36 |
+
"grad_norm": 0.4140625,
|
| 37 |
+
"learning_rate": 0.001,
|
| 38 |
+
"loss": 5.959280776977539,
|
| 39 |
+
"step": 60
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.0431266846361186,
|
| 43 |
+
"grad_norm": 0.2333984375,
|
| 44 |
+
"learning_rate": 0.001,
|
| 45 |
+
"loss": 5.698427963256836,
|
| 46 |
+
"step": 80
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.05390835579514825,
|
| 50 |
+
"grad_norm": 0.26953125,
|
| 51 |
+
"learning_rate": 0.001,
|
| 52 |
+
"loss": 5.483244323730469,
|
| 53 |
+
"step": 100
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.05390835579514825,
|
| 57 |
+
"eval_loss": 5.370211601257324,
|
| 58 |
+
"eval_runtime": 11.0192,
|
| 59 |
+
"eval_samples_per_second": 864.582,
|
| 60 |
+
"eval_steps_per_second": 1.724,
|
| 61 |
+
"step": 100
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.0646900269541779,
|
| 65 |
+
"grad_norm": 0.33203125,
|
| 66 |
+
"learning_rate": 0.001,
|
| 67 |
+
"loss": 5.284563827514648,
|
| 68 |
+
"step": 120
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.07547169811320754,
|
| 72 |
+
"grad_norm": 0.5859375,
|
| 73 |
+
"learning_rate": 0.001,
|
| 74 |
+
"loss": 5.074047470092774,
|
| 75 |
+
"step": 140
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.08086253369272237,
|
| 79 |
+
"eval_loss": 4.84706974029541,
|
| 80 |
+
"eval_runtime": 11.0423,
|
| 81 |
+
"eval_samples_per_second": 862.77,
|
| 82 |
+
"eval_steps_per_second": 1.721,
|
| 83 |
+
"step": 150
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.0862533692722372,
|
| 87 |
+
"grad_norm": 0.578125,
|
| 88 |
+
"learning_rate": 0.001,
|
| 89 |
+
"loss": 4.856560516357422,
|
| 90 |
+
"step": 160
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.09703504043126684,
|
| 94 |
+
"grad_norm": 0.76953125,
|
| 95 |
+
"learning_rate": 0.001,
|
| 96 |
+
"loss": 4.646049499511719,
|
| 97 |
+
"step": 180
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.1078167115902965,
|
| 101 |
+
"grad_norm": 0.32421875,
|
| 102 |
+
"learning_rate": 0.001,
|
| 103 |
+
"loss": 4.480390548706055,
|
| 104 |
+
"step": 200
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.1078167115902965,
|
| 108 |
+
"eval_loss": 4.395232677459717,
|
| 109 |
+
"eval_runtime": 11.0603,
|
| 110 |
+
"eval_samples_per_second": 861.369,
|
| 111 |
+
"eval_steps_per_second": 1.718,
|
| 112 |
+
"step": 200
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 0.11859838274932614,
|
| 116 |
+
"grad_norm": 0.57421875,
|
| 117 |
+
"learning_rate": 0.001,
|
| 118 |
+
"loss": 4.35283317565918,
|
| 119 |
+
"step": 220
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 0.1293800539083558,
|
| 123 |
+
"grad_norm": 0.64453125,
|
| 124 |
+
"learning_rate": 0.001,
|
| 125 |
+
"loss": 4.239774703979492,
|
| 126 |
+
"step": 240
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 0.1347708894878706,
|
| 130 |
+
"eval_loss": 4.130031108856201,
|
| 131 |
+
"eval_runtime": 11.5228,
|
| 132 |
+
"eval_samples_per_second": 826.793,
|
| 133 |
+
"eval_steps_per_second": 1.649,
|
| 134 |
+
"step": 250
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.14016172506738545,
|
| 138 |
+
"grad_norm": 1.3046875,
|
| 139 |
+
"learning_rate": 0.001,
|
| 140 |
+
"loss": 4.1399494171142575,
|
| 141 |
+
"step": 260
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.1509433962264151,
|
| 145 |
+
"grad_norm": 0.68359375,
|
| 146 |
+
"learning_rate": 0.001,
|
| 147 |
+
"loss": 4.071997833251953,
|
| 148 |
+
"step": 280
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 0.16172506738544473,
|
| 152 |
+
"grad_norm": 0.6484375,
|
| 153 |
+
"learning_rate": 0.001,
|
| 154 |
+
"loss": 4.0012977600097654,
|
| 155 |
+
"step": 300
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 0.16172506738544473,
|
| 159 |
+
"eval_loss": 3.970766067504883,
|
| 160 |
+
"eval_runtime": 11.2668,
|
| 161 |
+
"eval_samples_per_second": 845.58,
|
| 162 |
+
"eval_steps_per_second": 1.686,
|
| 163 |
+
"step": 300
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.1725067385444744,
|
| 167 |
+
"grad_norm": 0.494140625,
|
| 168 |
+
"learning_rate": 0.001,
|
| 169 |
+
"loss": 3.9426082611083983,
|
| 170 |
+
"step": 320
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.18328840970350405,
|
| 174 |
+
"grad_norm": 0.3359375,
|
| 175 |
+
"learning_rate": 0.001,
|
| 176 |
+
"loss": 3.884596252441406,
|
| 177 |
+
"step": 340
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.18867924528301888,
|
| 181 |
+
"eval_loss": 3.869180202484131,
|
| 182 |
+
"eval_runtime": 11.0104,
|
| 183 |
+
"eval_samples_per_second": 865.274,
|
| 184 |
+
"eval_steps_per_second": 1.726,
|
| 185 |
+
"step": 350
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.1940700808625337,
|
| 189 |
+
"grad_norm": 0.59765625,
|
| 190 |
+
"learning_rate": 0.001,
|
| 191 |
+
"loss": 3.846722412109375,
|
| 192 |
+
"step": 360
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.20485175202156333,
|
| 196 |
+
"grad_norm": 0.3203125,
|
| 197 |
+
"learning_rate": 0.001,
|
| 198 |
+
"loss": 3.7851001739501955,
|
| 199 |
+
"step": 380
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.215633423180593,
|
| 203 |
+
"grad_norm": 0.423828125,
|
| 204 |
+
"learning_rate": 0.001,
|
| 205 |
+
"loss": 3.758561706542969,
|
| 206 |
+
"step": 400
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.215633423180593,
|
| 210 |
+
"eval_loss": 3.7298622131347656,
|
| 211 |
+
"eval_runtime": 11.1,
|
| 212 |
+
"eval_samples_per_second": 858.29,
|
| 213 |
+
"eval_steps_per_second": 1.712,
|
| 214 |
+
"step": 400
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 0.22641509433962265,
|
| 218 |
+
"grad_norm": 0.330078125,
|
| 219 |
+
"learning_rate": 0.001,
|
| 220 |
+
"loss": 3.7036609649658203,
|
| 221 |
+
"step": 420
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.2371967654986523,
|
| 225 |
+
"grad_norm": 0.69140625,
|
| 226 |
+
"learning_rate": 0.001,
|
| 227 |
+
"loss": 3.6900096893310548,
|
| 228 |
+
"step": 440
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 0.24258760107816713,
|
| 232 |
+
"eval_loss": 3.649232864379883,
|
| 233 |
+
"eval_runtime": 11.3447,
|
| 234 |
+
"eval_samples_per_second": 839.777,
|
| 235 |
+
"eval_steps_per_second": 1.675,
|
| 236 |
+
"step": 450
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.24797843665768193,
|
| 240 |
+
"grad_norm": 0.42578125,
|
| 241 |
+
"learning_rate": 0.001,
|
| 242 |
+
"loss": 3.65367431640625,
|
| 243 |
+
"step": 460
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.2587601078167116,
|
| 247 |
+
"grad_norm": 0.70703125,
|
| 248 |
+
"learning_rate": 0.001,
|
| 249 |
+
"loss": 3.617287826538086,
|
| 250 |
+
"step": 480
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.2695417789757412,
|
| 254 |
+
"grad_norm": 0.68359375,
|
| 255 |
+
"learning_rate": 0.001,
|
| 256 |
+
"loss": 3.580358123779297,
|
| 257 |
+
"step": 500
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.2695417789757412,
|
| 261 |
+
"eval_loss": 3.590764284133911,
|
| 262 |
+
"eval_runtime": 11.1385,
|
| 263 |
+
"eval_samples_per_second": 855.32,
|
| 264 |
+
"eval_steps_per_second": 1.706,
|
| 265 |
+
"step": 500
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.2803234501347709,
|
| 269 |
+
"grad_norm": 0.373046875,
|
| 270 |
+
"learning_rate": 0.001,
|
| 271 |
+
"loss": 3.557994079589844,
|
| 272 |
+
"step": 520
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.29110512129380056,
|
| 276 |
+
"grad_norm": 0.671875,
|
| 277 |
+
"learning_rate": 0.001,
|
| 278 |
+
"loss": 3.5252918243408202,
|
| 279 |
+
"step": 540
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.29649595687331537,
|
| 283 |
+
"eval_loss": 3.510967969894409,
|
| 284 |
+
"eval_runtime": 11.0988,
|
| 285 |
+
"eval_samples_per_second": 858.384,
|
| 286 |
+
"eval_steps_per_second": 1.712,
|
| 287 |
+
"step": 550
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.3018867924528302,
|
| 291 |
+
"grad_norm": 0.34375,
|
| 292 |
+
"learning_rate": 0.001,
|
| 293 |
+
"loss": 3.5099483489990235,
|
| 294 |
+
"step": 560
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.31266846361185985,
|
| 298 |
+
"grad_norm": 0.77734375,
|
| 299 |
+
"learning_rate": 0.001,
|
| 300 |
+
"loss": 3.471272277832031,
|
| 301 |
+
"step": 580
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.32345013477088946,
|
| 305 |
+
"grad_norm": 0.62109375,
|
| 306 |
+
"learning_rate": 0.001,
|
| 307 |
+
"loss": 3.477389907836914,
|
| 308 |
+
"step": 600
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.32345013477088946,
|
| 312 |
+
"eval_loss": 3.446561098098755,
|
| 313 |
+
"eval_runtime": 11.1795,
|
| 314 |
+
"eval_samples_per_second": 852.182,
|
| 315 |
+
"eval_steps_per_second": 1.7,
|
| 316 |
+
"step": 600
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.33423180592991913,
|
| 320 |
+
"grad_norm": 0.453125,
|
| 321 |
+
"learning_rate": 0.001,
|
| 322 |
+
"loss": 3.427866744995117,
|
| 323 |
+
"step": 620
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 0.3450134770889488,
|
| 327 |
+
"grad_norm": 0.65234375,
|
| 328 |
+
"learning_rate": 0.001,
|
| 329 |
+
"loss": 3.4235225677490235,
|
| 330 |
+
"step": 640
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.3504043126684636,
|
| 334 |
+
"eval_loss": 3.397838830947876,
|
| 335 |
+
"eval_runtime": 11.492,
|
| 336 |
+
"eval_samples_per_second": 829.015,
|
| 337 |
+
"eval_steps_per_second": 1.653,
|
| 338 |
+
"step": 650
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.3557951482479784,
|
| 342 |
+
"grad_norm": 0.4140625,
|
| 343 |
+
"learning_rate": 0.001,
|
| 344 |
+
"loss": 3.4058517456054687,
|
| 345 |
+
"step": 660
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.3665768194070081,
|
| 349 |
+
"grad_norm": 1.0390625,
|
| 350 |
+
"learning_rate": 0.001,
|
| 351 |
+
"loss": 3.3697017669677733,
|
| 352 |
+
"step": 680
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.37735849056603776,
|
| 356 |
+
"grad_norm": 0.859375,
|
| 357 |
+
"learning_rate": 0.001,
|
| 358 |
+
"loss": 3.3665958404541017,
|
| 359 |
+
"step": 700
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.37735849056603776,
|
| 363 |
+
"eval_loss": 3.3444883823394775,
|
| 364 |
+
"eval_runtime": 11.5463,
|
| 365 |
+
"eval_samples_per_second": 825.113,
|
| 366 |
+
"eval_steps_per_second": 1.646,
|
| 367 |
+
"step": 700
|
| 368 |
+
}
|
| 369 |
+
],
|
| 370 |
+
"logging_steps": 20,
|
| 371 |
+
"max_steps": 1000,
|
| 372 |
+
"num_input_tokens_seen": 0,
|
| 373 |
+
"num_train_epochs": 1,
|
| 374 |
+
"save_steps": 100,
|
| 375 |
+
"stateful_callbacks": {
|
| 376 |
+
"TrainerControl": {
|
| 377 |
+
"args": {
|
| 378 |
+
"should_epoch_stop": false,
|
| 379 |
+
"should_evaluate": false,
|
| 380 |
+
"should_log": false,
|
| 381 |
+
"should_save": true,
|
| 382 |
+
"should_training_stop": false
|
| 383 |
+
},
|
| 384 |
+
"attributes": {}
|
| 385 |
+
}
|
| 386 |
+
},
|
| 387 |
+
"total_flos": 8310983177011200.0,
|
| 388 |
+
"train_batch_size": 512,
|
| 389 |
+
"trial_name": null,
|
| 390 |
+
"trial_params": null
|
| 391 |
+
}
|
out/mlp-sigmoid-46L_run/checkpoint-700/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f497915d25e5adfc7823f2b88d2f779dbbdb1cedd0353e013bfd915138617020
|
| 3 |
+
size 4920
|
out/mlp-sigmoid-46L_run/checkpoint-800/config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation": "sigmoid",
|
| 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": "mlp",
|
| 19 |
+
"model_type": "tiny_llama",
|
| 20 |
+
"num_attention_heads": 4,
|
| 21 |
+
"num_hidden_layers": 46,
|
| 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/mlp-sigmoid-46L_run/checkpoint-800/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dbcc7c32910da461d577ec24990c01d3027679fc3fe2ed2121a6d3668d18cb96
|
| 3 |
+
size 16186168
|
out/mlp-sigmoid-46L_run/checkpoint-800/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fc822863d570a85f697ed14cb15e4628dda71d56b6f4b30918bc9f414c29694e
|
| 3 |
+
size 32605416
|
out/mlp-sigmoid-46L_run/checkpoint-800/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f83fe17529e572dff2734bb21512b28dd7cf5d20ef0e84688f5068ffbf24e765
|
| 3 |
+
size 14244
|
out/mlp-sigmoid-46L_run/checkpoint-800/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7334026bbd250befa466dcbe6bec8bfe84c03024263e4dea55e40698f2caa92e
|
| 3 |
+
size 1064
|
out/mlp-sigmoid-46L_run/checkpoint-800/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/mlp-sigmoid-46L_run/checkpoint-800/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/mlp-sigmoid-46L_run/checkpoint-800/trainer_state.json
ADDED
|
@@ -0,0 +1,442 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.431266846361186,
|
| 6 |
+
"eval_steps": 50,
|
| 7 |
+
"global_step": 800,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01078167115902965,
|
| 14 |
+
"grad_norm": 0.953125,
|
| 15 |
+
"learning_rate": 0.001,
|
| 16 |
+
"loss": 7.132730865478516,
|
| 17 |
+
"step": 20
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0215633423180593,
|
| 21 |
+
"grad_norm": 0.30078125,
|
| 22 |
+
"learning_rate": 0.001,
|
| 23 |
+
"loss": 6.084678268432617,
|
| 24 |
+
"step": 40
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.026954177897574125,
|
| 28 |
+
"eval_loss": 5.9758501052856445,
|
| 29 |
+
"eval_runtime": 11.1232,
|
| 30 |
+
"eval_samples_per_second": 856.498,
|
| 31 |
+
"eval_steps_per_second": 1.708,
|
| 32 |
+
"step": 50
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.03234501347708895,
|
| 36 |
+
"grad_norm": 0.4140625,
|
| 37 |
+
"learning_rate": 0.001,
|
| 38 |
+
"loss": 5.959280776977539,
|
| 39 |
+
"step": 60
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.0431266846361186,
|
| 43 |
+
"grad_norm": 0.2333984375,
|
| 44 |
+
"learning_rate": 0.001,
|
| 45 |
+
"loss": 5.698427963256836,
|
| 46 |
+
"step": 80
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.05390835579514825,
|
| 50 |
+
"grad_norm": 0.26953125,
|
| 51 |
+
"learning_rate": 0.001,
|
| 52 |
+
"loss": 5.483244323730469,
|
| 53 |
+
"step": 100
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.05390835579514825,
|
| 57 |
+
"eval_loss": 5.370211601257324,
|
| 58 |
+
"eval_runtime": 11.0192,
|
| 59 |
+
"eval_samples_per_second": 864.582,
|
| 60 |
+
"eval_steps_per_second": 1.724,
|
| 61 |
+
"step": 100
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.0646900269541779,
|
| 65 |
+
"grad_norm": 0.33203125,
|
| 66 |
+
"learning_rate": 0.001,
|
| 67 |
+
"loss": 5.284563827514648,
|
| 68 |
+
"step": 120
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.07547169811320754,
|
| 72 |
+
"grad_norm": 0.5859375,
|
| 73 |
+
"learning_rate": 0.001,
|
| 74 |
+
"loss": 5.074047470092774,
|
| 75 |
+
"step": 140
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.08086253369272237,
|
| 79 |
+
"eval_loss": 4.84706974029541,
|
| 80 |
+
"eval_runtime": 11.0423,
|
| 81 |
+
"eval_samples_per_second": 862.77,
|
| 82 |
+
"eval_steps_per_second": 1.721,
|
| 83 |
+
"step": 150
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.0862533692722372,
|
| 87 |
+
"grad_norm": 0.578125,
|
| 88 |
+
"learning_rate": 0.001,
|
| 89 |
+
"loss": 4.856560516357422,
|
| 90 |
+
"step": 160
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.09703504043126684,
|
| 94 |
+
"grad_norm": 0.76953125,
|
| 95 |
+
"learning_rate": 0.001,
|
| 96 |
+
"loss": 4.646049499511719,
|
| 97 |
+
"step": 180
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.1078167115902965,
|
| 101 |
+
"grad_norm": 0.32421875,
|
| 102 |
+
"learning_rate": 0.001,
|
| 103 |
+
"loss": 4.480390548706055,
|
| 104 |
+
"step": 200
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.1078167115902965,
|
| 108 |
+
"eval_loss": 4.395232677459717,
|
| 109 |
+
"eval_runtime": 11.0603,
|
| 110 |
+
"eval_samples_per_second": 861.369,
|
| 111 |
+
"eval_steps_per_second": 1.718,
|
| 112 |
+
"step": 200
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 0.11859838274932614,
|
| 116 |
+
"grad_norm": 0.57421875,
|
| 117 |
+
"learning_rate": 0.001,
|
| 118 |
+
"loss": 4.35283317565918,
|
| 119 |
+
"step": 220
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 0.1293800539083558,
|
| 123 |
+
"grad_norm": 0.64453125,
|
| 124 |
+
"learning_rate": 0.001,
|
| 125 |
+
"loss": 4.239774703979492,
|
| 126 |
+
"step": 240
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 0.1347708894878706,
|
| 130 |
+
"eval_loss": 4.130031108856201,
|
| 131 |
+
"eval_runtime": 11.5228,
|
| 132 |
+
"eval_samples_per_second": 826.793,
|
| 133 |
+
"eval_steps_per_second": 1.649,
|
| 134 |
+
"step": 250
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.14016172506738545,
|
| 138 |
+
"grad_norm": 1.3046875,
|
| 139 |
+
"learning_rate": 0.001,
|
| 140 |
+
"loss": 4.1399494171142575,
|
| 141 |
+
"step": 260
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.1509433962264151,
|
| 145 |
+
"grad_norm": 0.68359375,
|
| 146 |
+
"learning_rate": 0.001,
|
| 147 |
+
"loss": 4.071997833251953,
|
| 148 |
+
"step": 280
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 0.16172506738544473,
|
| 152 |
+
"grad_norm": 0.6484375,
|
| 153 |
+
"learning_rate": 0.001,
|
| 154 |
+
"loss": 4.0012977600097654,
|
| 155 |
+
"step": 300
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 0.16172506738544473,
|
| 159 |
+
"eval_loss": 3.970766067504883,
|
| 160 |
+
"eval_runtime": 11.2668,
|
| 161 |
+
"eval_samples_per_second": 845.58,
|
| 162 |
+
"eval_steps_per_second": 1.686,
|
| 163 |
+
"step": 300
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.1725067385444744,
|
| 167 |
+
"grad_norm": 0.494140625,
|
| 168 |
+
"learning_rate": 0.001,
|
| 169 |
+
"loss": 3.9426082611083983,
|
| 170 |
+
"step": 320
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.18328840970350405,
|
| 174 |
+
"grad_norm": 0.3359375,
|
| 175 |
+
"learning_rate": 0.001,
|
| 176 |
+
"loss": 3.884596252441406,
|
| 177 |
+
"step": 340
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.18867924528301888,
|
| 181 |
+
"eval_loss": 3.869180202484131,
|
| 182 |
+
"eval_runtime": 11.0104,
|
| 183 |
+
"eval_samples_per_second": 865.274,
|
| 184 |
+
"eval_steps_per_second": 1.726,
|
| 185 |
+
"step": 350
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.1940700808625337,
|
| 189 |
+
"grad_norm": 0.59765625,
|
| 190 |
+
"learning_rate": 0.001,
|
| 191 |
+
"loss": 3.846722412109375,
|
| 192 |
+
"step": 360
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.20485175202156333,
|
| 196 |
+
"grad_norm": 0.3203125,
|
| 197 |
+
"learning_rate": 0.001,
|
| 198 |
+
"loss": 3.7851001739501955,
|
| 199 |
+
"step": 380
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.215633423180593,
|
| 203 |
+
"grad_norm": 0.423828125,
|
| 204 |
+
"learning_rate": 0.001,
|
| 205 |
+
"loss": 3.758561706542969,
|
| 206 |
+
"step": 400
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.215633423180593,
|
| 210 |
+
"eval_loss": 3.7298622131347656,
|
| 211 |
+
"eval_runtime": 11.1,
|
| 212 |
+
"eval_samples_per_second": 858.29,
|
| 213 |
+
"eval_steps_per_second": 1.712,
|
| 214 |
+
"step": 400
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 0.22641509433962265,
|
| 218 |
+
"grad_norm": 0.330078125,
|
| 219 |
+
"learning_rate": 0.001,
|
| 220 |
+
"loss": 3.7036609649658203,
|
| 221 |
+
"step": 420
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.2371967654986523,
|
| 225 |
+
"grad_norm": 0.69140625,
|
| 226 |
+
"learning_rate": 0.001,
|
| 227 |
+
"loss": 3.6900096893310548,
|
| 228 |
+
"step": 440
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 0.24258760107816713,
|
| 232 |
+
"eval_loss": 3.649232864379883,
|
| 233 |
+
"eval_runtime": 11.3447,
|
| 234 |
+
"eval_samples_per_second": 839.777,
|
| 235 |
+
"eval_steps_per_second": 1.675,
|
| 236 |
+
"step": 450
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.24797843665768193,
|
| 240 |
+
"grad_norm": 0.42578125,
|
| 241 |
+
"learning_rate": 0.001,
|
| 242 |
+
"loss": 3.65367431640625,
|
| 243 |
+
"step": 460
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.2587601078167116,
|
| 247 |
+
"grad_norm": 0.70703125,
|
| 248 |
+
"learning_rate": 0.001,
|
| 249 |
+
"loss": 3.617287826538086,
|
| 250 |
+
"step": 480
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.2695417789757412,
|
| 254 |
+
"grad_norm": 0.68359375,
|
| 255 |
+
"learning_rate": 0.001,
|
| 256 |
+
"loss": 3.580358123779297,
|
| 257 |
+
"step": 500
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.2695417789757412,
|
| 261 |
+
"eval_loss": 3.590764284133911,
|
| 262 |
+
"eval_runtime": 11.1385,
|
| 263 |
+
"eval_samples_per_second": 855.32,
|
| 264 |
+
"eval_steps_per_second": 1.706,
|
| 265 |
+
"step": 500
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.2803234501347709,
|
| 269 |
+
"grad_norm": 0.373046875,
|
| 270 |
+
"learning_rate": 0.001,
|
| 271 |
+
"loss": 3.557994079589844,
|
| 272 |
+
"step": 520
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.29110512129380056,
|
| 276 |
+
"grad_norm": 0.671875,
|
| 277 |
+
"learning_rate": 0.001,
|
| 278 |
+
"loss": 3.5252918243408202,
|
| 279 |
+
"step": 540
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.29649595687331537,
|
| 283 |
+
"eval_loss": 3.510967969894409,
|
| 284 |
+
"eval_runtime": 11.0988,
|
| 285 |
+
"eval_samples_per_second": 858.384,
|
| 286 |
+
"eval_steps_per_second": 1.712,
|
| 287 |
+
"step": 550
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.3018867924528302,
|
| 291 |
+
"grad_norm": 0.34375,
|
| 292 |
+
"learning_rate": 0.001,
|
| 293 |
+
"loss": 3.5099483489990235,
|
| 294 |
+
"step": 560
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.31266846361185985,
|
| 298 |
+
"grad_norm": 0.77734375,
|
| 299 |
+
"learning_rate": 0.001,
|
| 300 |
+
"loss": 3.471272277832031,
|
| 301 |
+
"step": 580
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.32345013477088946,
|
| 305 |
+
"grad_norm": 0.62109375,
|
| 306 |
+
"learning_rate": 0.001,
|
| 307 |
+
"loss": 3.477389907836914,
|
| 308 |
+
"step": 600
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.32345013477088946,
|
| 312 |
+
"eval_loss": 3.446561098098755,
|
| 313 |
+
"eval_runtime": 11.1795,
|
| 314 |
+
"eval_samples_per_second": 852.182,
|
| 315 |
+
"eval_steps_per_second": 1.7,
|
| 316 |
+
"step": 600
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.33423180592991913,
|
| 320 |
+
"grad_norm": 0.453125,
|
| 321 |
+
"learning_rate": 0.001,
|
| 322 |
+
"loss": 3.427866744995117,
|
| 323 |
+
"step": 620
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 0.3450134770889488,
|
| 327 |
+
"grad_norm": 0.65234375,
|
| 328 |
+
"learning_rate": 0.001,
|
| 329 |
+
"loss": 3.4235225677490235,
|
| 330 |
+
"step": 640
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.3504043126684636,
|
| 334 |
+
"eval_loss": 3.397838830947876,
|
| 335 |
+
"eval_runtime": 11.492,
|
| 336 |
+
"eval_samples_per_second": 829.015,
|
| 337 |
+
"eval_steps_per_second": 1.653,
|
| 338 |
+
"step": 650
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.3557951482479784,
|
| 342 |
+
"grad_norm": 0.4140625,
|
| 343 |
+
"learning_rate": 0.001,
|
| 344 |
+
"loss": 3.4058517456054687,
|
| 345 |
+
"step": 660
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.3665768194070081,
|
| 349 |
+
"grad_norm": 1.0390625,
|
| 350 |
+
"learning_rate": 0.001,
|
| 351 |
+
"loss": 3.3697017669677733,
|
| 352 |
+
"step": 680
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.37735849056603776,
|
| 356 |
+
"grad_norm": 0.859375,
|
| 357 |
+
"learning_rate": 0.001,
|
| 358 |
+
"loss": 3.3665958404541017,
|
| 359 |
+
"step": 700
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.37735849056603776,
|
| 363 |
+
"eval_loss": 3.3444883823394775,
|
| 364 |
+
"eval_runtime": 11.5463,
|
| 365 |
+
"eval_samples_per_second": 825.113,
|
| 366 |
+
"eval_steps_per_second": 1.646,
|
| 367 |
+
"step": 700
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.3881401617250674,
|
| 371 |
+
"grad_norm": 0.59375,
|
| 372 |
+
"learning_rate": 0.001,
|
| 373 |
+
"loss": 3.3385574340820314,
|
| 374 |
+
"step": 720
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.39892183288409705,
|
| 378 |
+
"grad_norm": 0.328125,
|
| 379 |
+
"learning_rate": 0.001,
|
| 380 |
+
"loss": 3.312589645385742,
|
| 381 |
+
"step": 740
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.40431266846361186,
|
| 385 |
+
"eval_loss": 3.295668840408325,
|
| 386 |
+
"eval_runtime": 11.457,
|
| 387 |
+
"eval_samples_per_second": 831.544,
|
| 388 |
+
"eval_steps_per_second": 1.658,
|
| 389 |
+
"step": 750
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.40970350404312667,
|
| 393 |
+
"grad_norm": 1.015625,
|
| 394 |
+
"learning_rate": 0.001,
|
| 395 |
+
"loss": 3.3023757934570312,
|
| 396 |
+
"step": 760
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.42048517520215634,
|
| 400 |
+
"grad_norm": 0.515625,
|
| 401 |
+
"learning_rate": 0.001,
|
| 402 |
+
"loss": 3.3048343658447266,
|
| 403 |
+
"step": 780
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.431266846361186,
|
| 407 |
+
"grad_norm": 0.78125,
|
| 408 |
+
"learning_rate": 0.001,
|
| 409 |
+
"loss": 3.2649677276611326,
|
| 410 |
+
"step": 800
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.431266846361186,
|
| 414 |
+
"eval_loss": 3.2709333896636963,
|
| 415 |
+
"eval_runtime": 11.2527,
|
| 416 |
+
"eval_samples_per_second": 846.644,
|
| 417 |
+
"eval_steps_per_second": 1.688,
|
| 418 |
+
"step": 800
|
| 419 |
+
}
|
| 420 |
+
],
|
| 421 |
+
"logging_steps": 20,
|
| 422 |
+
"max_steps": 1000,
|
| 423 |
+
"num_input_tokens_seen": 0,
|
| 424 |
+
"num_train_epochs": 1,
|
| 425 |
+
"save_steps": 100,
|
| 426 |
+
"stateful_callbacks": {
|
| 427 |
+
"TrainerControl": {
|
| 428 |
+
"args": {
|
| 429 |
+
"should_epoch_stop": false,
|
| 430 |
+
"should_evaluate": false,
|
| 431 |
+
"should_log": false,
|
| 432 |
+
"should_save": true,
|
| 433 |
+
"should_training_stop": false
|
| 434 |
+
},
|
| 435 |
+
"attributes": {}
|
| 436 |
+
}
|
| 437 |
+
},
|
| 438 |
+
"total_flos": 9498266488012800.0,
|
| 439 |
+
"train_batch_size": 512,
|
| 440 |
+
"trial_name": null,
|
| 441 |
+
"trial_params": null
|
| 442 |
+
}
|
out/mlp-sigmoid-46L_run/checkpoint-800/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f497915d25e5adfc7823f2b88d2f779dbbdb1cedd0353e013bfd915138617020
|
| 3 |
+
size 4920
|
out/mlp-sigmoid-46L_run/checkpoint-900/config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation": "sigmoid",
|
| 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": "mlp",
|
| 19 |
+
"model_type": "tiny_llama",
|
| 20 |
+
"num_attention_heads": 4,
|
| 21 |
+
"num_hidden_layers": 46,
|
| 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/mlp-sigmoid-46L_run/checkpoint-900/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:68a551015537011253b4c5d39afde90565582343a162faeeafc880d97a9aa0e7
|
| 3 |
+
size 16186168
|
out/mlp-sigmoid-46L_run/checkpoint-900/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:17291c3c1fdc1236b18fb25da1e9bb9d52d79c3d1db29e22126afee51792c5d8
|
| 3 |
+
size 32605416
|
out/mlp-sigmoid-46L_run/checkpoint-900/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:86a2aa7bca84d2b8e7dd4e04a286714ba9169af11c46c739950a52df4c45259f
|
| 3 |
+
size 14244
|
out/mlp-sigmoid-46L_run/checkpoint-900/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6d48b00fea613f19335b87fb622648d28d7f8900963d07ba43c75df2b4dc3d4e
|
| 3 |
+
size 1064
|
out/mlp-sigmoid-46L_run/checkpoint-900/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/mlp-sigmoid-46L_run/checkpoint-900/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 |
+
}
|