Training in progress, step 400, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 859942080
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ee5b1434a373cdcc5dad1a7067047ecfac41dd752c5091ac03443cfcd45e757f
|
| 3 |
size 859942080
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 90187222
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b25555b38342e1ba69cdf11c523aea0de6238361604dd97d0958ab7cd0805b0d
|
| 3 |
size 90187222
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6bede8cf68b594f047239405eb697b986cf2eee2f6054fce463ec7f666ac8d8a
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2080
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:92be4570eef35209dec51037fce376384799ea8aa047398b6b8d22e11ae0faf5
|
| 3 |
size 2080
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 1.802311658859253,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-100",
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 100,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -2139,6 +2139,714 @@
|
|
| 2139 |
"eval_samples_per_second": 5.711,
|
| 2140 |
"eval_steps_per_second": 1.904,
|
| 2141 |
"step": 300
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2142 |
}
|
| 2143 |
],
|
| 2144 |
"logging_steps": 1,
|
|
@@ -2153,7 +2861,7 @@
|
|
| 2153 |
"early_stopping_threshold": 0.0
|
| 2154 |
},
|
| 2155 |
"attributes": {
|
| 2156 |
-
"early_stopping_patience_counter":
|
| 2157 |
}
|
| 2158 |
},
|
| 2159 |
"TrainerControl": {
|
|
@@ -2162,12 +2870,12 @@
|
|
| 2162 |
"should_evaluate": false,
|
| 2163 |
"should_log": false,
|
| 2164 |
"should_save": true,
|
| 2165 |
-
"should_training_stop":
|
| 2166 |
},
|
| 2167 |
"attributes": {}
|
| 2168 |
}
|
| 2169 |
},
|
| 2170 |
-
"total_flos":
|
| 2171 |
"train_batch_size": 3,
|
| 2172 |
"trial_name": null,
|
| 2173 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 1.802311658859253,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-100",
|
| 4 |
+
"epoch": 0.0029126708463493313,
|
| 5 |
"eval_steps": 100,
|
| 6 |
+
"global_step": 400,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2139 |
"eval_samples_per_second": 5.711,
|
| 2140 |
"eval_steps_per_second": 1.904,
|
| 2141 |
"step": 300
|
| 2142 |
+
},
|
| 2143 |
+
{
|
| 2144 |
+
"epoch": 0.002191784811877872,
|
| 2145 |
+
"grad_norm": 8.730939865112305,
|
| 2146 |
+
"learning_rate": 0.0001999997766688466,
|
| 2147 |
+
"loss": 1.4397,
|
| 2148 |
+
"step": 301
|
| 2149 |
+
},
|
| 2150 |
+
{
|
| 2151 |
+
"epoch": 0.002199066488993745,
|
| 2152 |
+
"grad_norm": 11.153186798095703,
|
| 2153 |
+
"learning_rate": 0.0001999997766688466,
|
| 2154 |
+
"loss": 1.4227,
|
| 2155 |
+
"step": 302
|
| 2156 |
+
},
|
| 2157 |
+
{
|
| 2158 |
+
"epoch": 0.002206348166109618,
|
| 2159 |
+
"grad_norm": 17.12929344177246,
|
| 2160 |
+
"learning_rate": 0.0001999997766688466,
|
| 2161 |
+
"loss": 2.2076,
|
| 2162 |
+
"step": 303
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 0.0022136298432254915,
|
| 2166 |
+
"grad_norm": 25.37128257751465,
|
| 2167 |
+
"learning_rate": 0.0001999997766688466,
|
| 2168 |
+
"loss": 1.7579,
|
| 2169 |
+
"step": 304
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 0.002220911520341365,
|
| 2173 |
+
"grad_norm": 12.106339454650879,
|
| 2174 |
+
"learning_rate": 0.0001999997766688466,
|
| 2175 |
+
"loss": 1.8253,
|
| 2176 |
+
"step": 305
|
| 2177 |
+
},
|
| 2178 |
+
{
|
| 2179 |
+
"epoch": 0.0022281931974572383,
|
| 2180 |
+
"grad_norm": 27.969057083129883,
|
| 2181 |
+
"learning_rate": 0.00019999976211693138,
|
| 2182 |
+
"loss": 1.8772,
|
| 2183 |
+
"step": 306
|
| 2184 |
+
},
|
| 2185 |
+
{
|
| 2186 |
+
"epoch": 0.0022354748745731117,
|
| 2187 |
+
"grad_norm": 22.067020416259766,
|
| 2188 |
+
"learning_rate": 0.00019999976211693138,
|
| 2189 |
+
"loss": 2.8781,
|
| 2190 |
+
"step": 307
|
| 2191 |
+
},
|
| 2192 |
+
{
|
| 2193 |
+
"epoch": 0.002242756551688985,
|
| 2194 |
+
"grad_norm": 9.046490669250488,
|
| 2195 |
+
"learning_rate": 0.00019999976211693138,
|
| 2196 |
+
"loss": 1.531,
|
| 2197 |
+
"step": 308
|
| 2198 |
+
},
|
| 2199 |
+
{
|
| 2200 |
+
"epoch": 0.0022500382288048585,
|
| 2201 |
+
"grad_norm": 7.611325740814209,
|
| 2202 |
+
"learning_rate": 0.00019999976211693138,
|
| 2203 |
+
"loss": 1.7938,
|
| 2204 |
+
"step": 309
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 0.0022573199059207315,
|
| 2208 |
+
"grad_norm": 14.951436996459961,
|
| 2209 |
+
"learning_rate": 0.00019999976211693138,
|
| 2210 |
+
"loss": 1.4902,
|
| 2211 |
+
"step": 310
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 0.002264601583036605,
|
| 2215 |
+
"grad_norm": 15.602734565734863,
|
| 2216 |
+
"learning_rate": 0.00019999976211693138,
|
| 2217 |
+
"loss": 1.874,
|
| 2218 |
+
"step": 311
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"epoch": 0.0022718832601524783,
|
| 2222 |
+
"grad_norm": 8.054279327392578,
|
| 2223 |
+
"learning_rate": 0.00019999976211693138,
|
| 2224 |
+
"loss": 1.3136,
|
| 2225 |
+
"step": 312
|
| 2226 |
+
},
|
| 2227 |
+
{
|
| 2228 |
+
"epoch": 0.0022791649372683517,
|
| 2229 |
+
"grad_norm": 7.248758316040039,
|
| 2230 |
+
"learning_rate": 0.00019999976211693138,
|
| 2231 |
+
"loss": 2.018,
|
| 2232 |
+
"step": 313
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 0.002286446614384225,
|
| 2236 |
+
"grad_norm": 13.146790504455566,
|
| 2237 |
+
"learning_rate": 0.00019999976211693138,
|
| 2238 |
+
"loss": 2.1679,
|
| 2239 |
+
"step": 314
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 0.0022937282915000985,
|
| 2243 |
+
"grad_norm": 13.931244850158691,
|
| 2244 |
+
"learning_rate": 0.00019999976211693138,
|
| 2245 |
+
"loss": 2.2164,
|
| 2246 |
+
"step": 315
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 0.002301009968615972,
|
| 2250 |
+
"grad_norm": 15.501871109008789,
|
| 2251 |
+
"learning_rate": 0.00019999976211693138,
|
| 2252 |
+
"loss": 2.2128,
|
| 2253 |
+
"step": 316
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 0.002308291645731845,
|
| 2257 |
+
"grad_norm": 11.91884708404541,
|
| 2258 |
+
"learning_rate": 0.00019999976211693138,
|
| 2259 |
+
"loss": 1.908,
|
| 2260 |
+
"step": 317
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 0.002315573322847718,
|
| 2264 |
+
"grad_norm": 8.88620662689209,
|
| 2265 |
+
"learning_rate": 0.00019999976211693138,
|
| 2266 |
+
"loss": 1.7827,
|
| 2267 |
+
"step": 318
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 0.0023228549999635916,
|
| 2271 |
+
"grad_norm": 8.506556510925293,
|
| 2272 |
+
"learning_rate": 0.00019999976211693138,
|
| 2273 |
+
"loss": 1.6667,
|
| 2274 |
+
"step": 319
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 0.002330136677079465,
|
| 2278 |
+
"grad_norm": 7.398971080780029,
|
| 2279 |
+
"learning_rate": 0.00019999974756501615,
|
| 2280 |
+
"loss": 1.7337,
|
| 2281 |
+
"step": 320
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 0.0023374183541953384,
|
| 2285 |
+
"grad_norm": 14.629541397094727,
|
| 2286 |
+
"learning_rate": 0.00019999974756501615,
|
| 2287 |
+
"loss": 2.8259,
|
| 2288 |
+
"step": 321
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 0.0023447000313112118,
|
| 2292 |
+
"grad_norm": 10.868115425109863,
|
| 2293 |
+
"learning_rate": 0.00019999974756501615,
|
| 2294 |
+
"loss": 1.6023,
|
| 2295 |
+
"step": 322
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 0.0023519817084270847,
|
| 2299 |
+
"grad_norm": 6.996133327484131,
|
| 2300 |
+
"learning_rate": 0.00019999974756501615,
|
| 2301 |
+
"loss": 1.479,
|
| 2302 |
+
"step": 323
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 0.002359263385542958,
|
| 2306 |
+
"grad_norm": 12.882672309875488,
|
| 2307 |
+
"learning_rate": 0.00019999974756501615,
|
| 2308 |
+
"loss": 1.9858,
|
| 2309 |
+
"step": 324
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 0.0023665450626588315,
|
| 2313 |
+
"grad_norm": 15.347146987915039,
|
| 2314 |
+
"learning_rate": 0.00019999974756501615,
|
| 2315 |
+
"loss": 2.4452,
|
| 2316 |
+
"step": 325
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 0.002373826739774705,
|
| 2320 |
+
"grad_norm": 12.391621589660645,
|
| 2321 |
+
"learning_rate": 0.00019999974756501615,
|
| 2322 |
+
"loss": 1.6006,
|
| 2323 |
+
"step": 326
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 0.0023811084168905783,
|
| 2327 |
+
"grad_norm": 9.05659294128418,
|
| 2328 |
+
"learning_rate": 0.00019999973301310092,
|
| 2329 |
+
"loss": 1.5926,
|
| 2330 |
+
"step": 327
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 0.0023883900940064517,
|
| 2334 |
+
"grad_norm": 17.410442352294922,
|
| 2335 |
+
"learning_rate": 0.00019999973301310092,
|
| 2336 |
+
"loss": 1.8012,
|
| 2337 |
+
"step": 328
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 0.002395671771122325,
|
| 2341 |
+
"grad_norm": 10.399604797363281,
|
| 2342 |
+
"learning_rate": 0.00019999973301310092,
|
| 2343 |
+
"loss": 1.7668,
|
| 2344 |
+
"step": 329
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 0.002402953448238198,
|
| 2348 |
+
"grad_norm": 10.19865894317627,
|
| 2349 |
+
"learning_rate": 0.00019999973301310092,
|
| 2350 |
+
"loss": 1.2579,
|
| 2351 |
+
"step": 330
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 0.0024102351253540715,
|
| 2355 |
+
"grad_norm": 11.40501594543457,
|
| 2356 |
+
"learning_rate": 0.00019999973301310092,
|
| 2357 |
+
"loss": 2.3442,
|
| 2358 |
+
"step": 331
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 0.002417516802469945,
|
| 2362 |
+
"grad_norm": 157.9639892578125,
|
| 2363 |
+
"learning_rate": 0.00019999973301310092,
|
| 2364 |
+
"loss": 2.4881,
|
| 2365 |
+
"step": 332
|
| 2366 |
+
},
|
| 2367 |
+
{
|
| 2368 |
+
"epoch": 0.0024247984795858183,
|
| 2369 |
+
"grad_norm": 38.14295196533203,
|
| 2370 |
+
"learning_rate": 0.00019999973301310092,
|
| 2371 |
+
"loss": 1.6426,
|
| 2372 |
+
"step": 333
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 0.0024320801567016916,
|
| 2376 |
+
"grad_norm": 81.4266357421875,
|
| 2377 |
+
"learning_rate": 0.0001999997184611857,
|
| 2378 |
+
"loss": 2.0941,
|
| 2379 |
+
"step": 334
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 0.002439361833817565,
|
| 2383 |
+
"grad_norm": 543.3828735351562,
|
| 2384 |
+
"learning_rate": 0.0001999997184611857,
|
| 2385 |
+
"loss": 3.5042,
|
| 2386 |
+
"step": 335
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 0.002446643510933438,
|
| 2390 |
+
"grad_norm": 136.7288360595703,
|
| 2391 |
+
"learning_rate": 0.0001999997184611857,
|
| 2392 |
+
"loss": 3.6408,
|
| 2393 |
+
"step": 336
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 0.0024539251880493114,
|
| 2397 |
+
"grad_norm": 351.3304443359375,
|
| 2398 |
+
"learning_rate": 0.0001999997184611857,
|
| 2399 |
+
"loss": 4.8935,
|
| 2400 |
+
"step": 337
|
| 2401 |
+
},
|
| 2402 |
+
{
|
| 2403 |
+
"epoch": 0.002461206865165185,
|
| 2404 |
+
"grad_norm": 25.63802146911621,
|
| 2405 |
+
"learning_rate": 0.0001999997184611857,
|
| 2406 |
+
"loss": 1.973,
|
| 2407 |
+
"step": 338
|
| 2408 |
+
},
|
| 2409 |
+
{
|
| 2410 |
+
"epoch": 0.002468488542281058,
|
| 2411 |
+
"grad_norm": 944.34716796875,
|
| 2412 |
+
"learning_rate": 0.0001999997184611857,
|
| 2413 |
+
"loss": 3.7135,
|
| 2414 |
+
"step": 339
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 0.0024757702193969316,
|
| 2418 |
+
"grad_norm": 53.539249420166016,
|
| 2419 |
+
"learning_rate": 0.0001999997184611857,
|
| 2420 |
+
"loss": 2.938,
|
| 2421 |
+
"step": 340
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 0.002483051896512805,
|
| 2425 |
+
"grad_norm": 30.831520080566406,
|
| 2426 |
+
"learning_rate": 0.00019999970390927047,
|
| 2427 |
+
"loss": 1.9902,
|
| 2428 |
+
"step": 341
|
| 2429 |
+
},
|
| 2430 |
+
{
|
| 2431 |
+
"epoch": 0.0024903335736286784,
|
| 2432 |
+
"grad_norm": 22.046127319335938,
|
| 2433 |
+
"learning_rate": 0.00019999970390927047,
|
| 2434 |
+
"loss": 2.3988,
|
| 2435 |
+
"step": 342
|
| 2436 |
+
},
|
| 2437 |
+
{
|
| 2438 |
+
"epoch": 0.0024976152507445513,
|
| 2439 |
+
"grad_norm": 10.293771743774414,
|
| 2440 |
+
"learning_rate": 0.00019999970390927047,
|
| 2441 |
+
"loss": 2.0854,
|
| 2442 |
+
"step": 343
|
| 2443 |
+
},
|
| 2444 |
+
{
|
| 2445 |
+
"epoch": 0.0025048969278604247,
|
| 2446 |
+
"grad_norm": 9.61192512512207,
|
| 2447 |
+
"learning_rate": 0.00019999970390927047,
|
| 2448 |
+
"loss": 1.9325,
|
| 2449 |
+
"step": 344
|
| 2450 |
+
},
|
| 2451 |
+
{
|
| 2452 |
+
"epoch": 0.002512178604976298,
|
| 2453 |
+
"grad_norm": 44.886131286621094,
|
| 2454 |
+
"learning_rate": 0.00019999970390927047,
|
| 2455 |
+
"loss": 2.0442,
|
| 2456 |
+
"step": 345
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 0.0025194602820921715,
|
| 2460 |
+
"grad_norm": 68.03099060058594,
|
| 2461 |
+
"learning_rate": 0.00019999970390927047,
|
| 2462 |
+
"loss": 2.3183,
|
| 2463 |
+
"step": 346
|
| 2464 |
+
},
|
| 2465 |
+
{
|
| 2466 |
+
"epoch": 0.002526741959208045,
|
| 2467 |
+
"grad_norm": 10.313465118408203,
|
| 2468 |
+
"learning_rate": 0.00019999970390927047,
|
| 2469 |
+
"loss": 1.682,
|
| 2470 |
+
"step": 347
|
| 2471 |
+
},
|
| 2472 |
+
{
|
| 2473 |
+
"epoch": 0.0025340236363239183,
|
| 2474 |
+
"grad_norm": 8.795594215393066,
|
| 2475 |
+
"learning_rate": 0.00019999970390927047,
|
| 2476 |
+
"loss": 1.7499,
|
| 2477 |
+
"step": 348
|
| 2478 |
+
},
|
| 2479 |
+
{
|
| 2480 |
+
"epoch": 0.0025413053134397913,
|
| 2481 |
+
"grad_norm": 15.358119010925293,
|
| 2482 |
+
"learning_rate": 0.00019999970390927047,
|
| 2483 |
+
"loss": 2.5658,
|
| 2484 |
+
"step": 349
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 0.0025485869905556647,
|
| 2488 |
+
"grad_norm": 13.6449556350708,
|
| 2489 |
+
"learning_rate": 0.00019999970390927047,
|
| 2490 |
+
"loss": 2.0518,
|
| 2491 |
+
"step": 350
|
| 2492 |
+
},
|
| 2493 |
+
{
|
| 2494 |
+
"epoch": 0.002555868667671538,
|
| 2495 |
+
"grad_norm": 9.879755020141602,
|
| 2496 |
+
"learning_rate": 0.00019999970390927047,
|
| 2497 |
+
"loss": 2.0147,
|
| 2498 |
+
"step": 351
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 0.0025631503447874114,
|
| 2502 |
+
"grad_norm": 36.2411003112793,
|
| 2503 |
+
"learning_rate": 0.00019999970390927047,
|
| 2504 |
+
"loss": 1.908,
|
| 2505 |
+
"step": 352
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 0.002570432021903285,
|
| 2509 |
+
"grad_norm": 10.988778114318848,
|
| 2510 |
+
"learning_rate": 0.00019999970390927047,
|
| 2511 |
+
"loss": 1.9842,
|
| 2512 |
+
"step": 353
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 0.0025777136990191582,
|
| 2516 |
+
"grad_norm": 46.944091796875,
|
| 2517 |
+
"learning_rate": 0.00019999968935735524,
|
| 2518 |
+
"loss": 1.9198,
|
| 2519 |
+
"step": 354
|
| 2520 |
+
},
|
| 2521 |
+
{
|
| 2522 |
+
"epoch": 0.0025849953761350316,
|
| 2523 |
+
"grad_norm": 136.6161651611328,
|
| 2524 |
+
"learning_rate": 0.00019999968935735524,
|
| 2525 |
+
"loss": 6.4679,
|
| 2526 |
+
"step": 355
|
| 2527 |
+
},
|
| 2528 |
+
{
|
| 2529 |
+
"epoch": 0.0025922770532509046,
|
| 2530 |
+
"grad_norm": 10.581056594848633,
|
| 2531 |
+
"learning_rate": 0.00019999968935735524,
|
| 2532 |
+
"loss": 1.8126,
|
| 2533 |
+
"step": 356
|
| 2534 |
+
},
|
| 2535 |
+
{
|
| 2536 |
+
"epoch": 0.002599558730366778,
|
| 2537 |
+
"grad_norm": 13.883180618286133,
|
| 2538 |
+
"learning_rate": 0.00019999968935735524,
|
| 2539 |
+
"loss": 1.4649,
|
| 2540 |
+
"step": 357
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 0.0026068404074826514,
|
| 2544 |
+
"grad_norm": 10.422268867492676,
|
| 2545 |
+
"learning_rate": 0.00019999968935735524,
|
| 2546 |
+
"loss": 2.4318,
|
| 2547 |
+
"step": 358
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 0.0026141220845985248,
|
| 2551 |
+
"grad_norm": 7.1867194175720215,
|
| 2552 |
+
"learning_rate": 0.00019999968935735524,
|
| 2553 |
+
"loss": 1.2992,
|
| 2554 |
+
"step": 359
|
| 2555 |
+
},
|
| 2556 |
+
{
|
| 2557 |
+
"epoch": 0.002621403761714398,
|
| 2558 |
+
"grad_norm": 10.631929397583008,
|
| 2559 |
+
"learning_rate": 0.00019999967480544,
|
| 2560 |
+
"loss": 1.3248,
|
| 2561 |
+
"step": 360
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 0.0026286854388302716,
|
| 2565 |
+
"grad_norm": 5.5126471519470215,
|
| 2566 |
+
"learning_rate": 0.00019999967480544,
|
| 2567 |
+
"loss": 1.244,
|
| 2568 |
+
"step": 361
|
| 2569 |
+
},
|
| 2570 |
+
{
|
| 2571 |
+
"epoch": 0.0026359671159461445,
|
| 2572 |
+
"grad_norm": 13.727178573608398,
|
| 2573 |
+
"learning_rate": 0.00019999967480544,
|
| 2574 |
+
"loss": 1.7372,
|
| 2575 |
+
"step": 362
|
| 2576 |
+
},
|
| 2577 |
+
{
|
| 2578 |
+
"epoch": 0.002643248793062018,
|
| 2579 |
+
"grad_norm": 11.092156410217285,
|
| 2580 |
+
"learning_rate": 0.00019999967480544,
|
| 2581 |
+
"loss": 1.9153,
|
| 2582 |
+
"step": 363
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 0.0026505304701778913,
|
| 2586 |
+
"grad_norm": 9.164811134338379,
|
| 2587 |
+
"learning_rate": 0.00019999967480544,
|
| 2588 |
+
"loss": 2.42,
|
| 2589 |
+
"step": 364
|
| 2590 |
+
},
|
| 2591 |
+
{
|
| 2592 |
+
"epoch": 0.0026578121472937647,
|
| 2593 |
+
"grad_norm": 9.590070724487305,
|
| 2594 |
+
"learning_rate": 0.00019999967480544,
|
| 2595 |
+
"loss": 2.405,
|
| 2596 |
+
"step": 365
|
| 2597 |
+
},
|
| 2598 |
+
{
|
| 2599 |
+
"epoch": 0.002665093824409638,
|
| 2600 |
+
"grad_norm": 15.919793128967285,
|
| 2601 |
+
"learning_rate": 0.00019999966025352478,
|
| 2602 |
+
"loss": 2.6192,
|
| 2603 |
+
"step": 366
|
| 2604 |
+
},
|
| 2605 |
+
{
|
| 2606 |
+
"epoch": 0.0026723755015255115,
|
| 2607 |
+
"grad_norm": 13.85263442993164,
|
| 2608 |
+
"learning_rate": 0.00019999966025352478,
|
| 2609 |
+
"loss": 2.0706,
|
| 2610 |
+
"step": 367
|
| 2611 |
+
},
|
| 2612 |
+
{
|
| 2613 |
+
"epoch": 0.002679657178641385,
|
| 2614 |
+
"grad_norm": 8.932114601135254,
|
| 2615 |
+
"learning_rate": 0.00019999966025352478,
|
| 2616 |
+
"loss": 1.2626,
|
| 2617 |
+
"step": 368
|
| 2618 |
+
},
|
| 2619 |
+
{
|
| 2620 |
+
"epoch": 0.002686938855757258,
|
| 2621 |
+
"grad_norm": 13.381108283996582,
|
| 2622 |
+
"learning_rate": 0.00019999966025352478,
|
| 2623 |
+
"loss": 2.2716,
|
| 2624 |
+
"step": 369
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 0.0026942205328731312,
|
| 2628 |
+
"grad_norm": 10.97152042388916,
|
| 2629 |
+
"learning_rate": 0.00019999966025352478,
|
| 2630 |
+
"loss": 2.5749,
|
| 2631 |
+
"step": 370
|
| 2632 |
+
},
|
| 2633 |
+
{
|
| 2634 |
+
"epoch": 0.0027015022099890046,
|
| 2635 |
+
"grad_norm": 11.363154411315918,
|
| 2636 |
+
"learning_rate": 0.00019999966025352478,
|
| 2637 |
+
"loss": 1.6513,
|
| 2638 |
+
"step": 371
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"epoch": 0.002708783887104878,
|
| 2642 |
+
"grad_norm": 10.766716003417969,
|
| 2643 |
+
"learning_rate": 0.00019999964570160955,
|
| 2644 |
+
"loss": 1.9178,
|
| 2645 |
+
"step": 372
|
| 2646 |
+
},
|
| 2647 |
+
{
|
| 2648 |
+
"epoch": 0.0027160655642207514,
|
| 2649 |
+
"grad_norm": 13.253247261047363,
|
| 2650 |
+
"learning_rate": 0.00019999964570160955,
|
| 2651 |
+
"loss": 1.9283,
|
| 2652 |
+
"step": 373
|
| 2653 |
+
},
|
| 2654 |
+
{
|
| 2655 |
+
"epoch": 0.002723347241336625,
|
| 2656 |
+
"grad_norm": 10.64928150177002,
|
| 2657 |
+
"learning_rate": 0.00019999964570160955,
|
| 2658 |
+
"loss": 1.8352,
|
| 2659 |
+
"step": 374
|
| 2660 |
+
},
|
| 2661 |
+
{
|
| 2662 |
+
"epoch": 0.002730628918452498,
|
| 2663 |
+
"grad_norm": 9.825387954711914,
|
| 2664 |
+
"learning_rate": 0.00019999964570160955,
|
| 2665 |
+
"loss": 1.9724,
|
| 2666 |
+
"step": 375
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 0.002737910595568371,
|
| 2670 |
+
"grad_norm": 9.247831344604492,
|
| 2671 |
+
"learning_rate": 0.00019999964570160955,
|
| 2672 |
+
"loss": 1.3778,
|
| 2673 |
+
"step": 376
|
| 2674 |
+
},
|
| 2675 |
+
{
|
| 2676 |
+
"epoch": 0.0027451922726842446,
|
| 2677 |
+
"grad_norm": 32.254974365234375,
|
| 2678 |
+
"learning_rate": 0.00019999964570160955,
|
| 2679 |
+
"loss": 1.9512,
|
| 2680 |
+
"step": 377
|
| 2681 |
+
},
|
| 2682 |
+
{
|
| 2683 |
+
"epoch": 0.002752473949800118,
|
| 2684 |
+
"grad_norm": 10.275064468383789,
|
| 2685 |
+
"learning_rate": 0.00019999963114969432,
|
| 2686 |
+
"loss": 1.8973,
|
| 2687 |
+
"step": 378
|
| 2688 |
+
},
|
| 2689 |
+
{
|
| 2690 |
+
"epoch": 0.0027597556269159914,
|
| 2691 |
+
"grad_norm": 10.921187400817871,
|
| 2692 |
+
"learning_rate": 0.00019999963114969432,
|
| 2693 |
+
"loss": 1.935,
|
| 2694 |
+
"step": 379
|
| 2695 |
+
},
|
| 2696 |
+
{
|
| 2697 |
+
"epoch": 0.0027670373040318648,
|
| 2698 |
+
"grad_norm": 8.636027336120605,
|
| 2699 |
+
"learning_rate": 0.00019999963114969432,
|
| 2700 |
+
"loss": 1.7587,
|
| 2701 |
+
"step": 380
|
| 2702 |
+
},
|
| 2703 |
+
{
|
| 2704 |
+
"epoch": 0.002774318981147738,
|
| 2705 |
+
"grad_norm": 12.754554748535156,
|
| 2706 |
+
"learning_rate": 0.00019999963114969432,
|
| 2707 |
+
"loss": 1.3819,
|
| 2708 |
+
"step": 381
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 0.002781600658263611,
|
| 2712 |
+
"grad_norm": 12.138449668884277,
|
| 2713 |
+
"learning_rate": 0.00019999963114969432,
|
| 2714 |
+
"loss": 1.4523,
|
| 2715 |
+
"step": 382
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 0.0027888823353794845,
|
| 2719 |
+
"grad_norm": 10.390461921691895,
|
| 2720 |
+
"learning_rate": 0.00019999963114969432,
|
| 2721 |
+
"loss": 1.6211,
|
| 2722 |
+
"step": 383
|
| 2723 |
+
},
|
| 2724 |
+
{
|
| 2725 |
+
"epoch": 0.002796164012495358,
|
| 2726 |
+
"grad_norm": 16.587440490722656,
|
| 2727 |
+
"learning_rate": 0.00019999963114969432,
|
| 2728 |
+
"loss": 2.1905,
|
| 2729 |
+
"step": 384
|
| 2730 |
+
},
|
| 2731 |
+
{
|
| 2732 |
+
"epoch": 0.0028034456896112313,
|
| 2733 |
+
"grad_norm": 15.672696113586426,
|
| 2734 |
+
"learning_rate": 0.00019999963114969432,
|
| 2735 |
+
"loss": 1.5341,
|
| 2736 |
+
"step": 385
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 0.0028107273667271047,
|
| 2740 |
+
"grad_norm": 12.346440315246582,
|
| 2741 |
+
"learning_rate": 0.00019999963114969432,
|
| 2742 |
+
"loss": 2.43,
|
| 2743 |
+
"step": 386
|
| 2744 |
+
},
|
| 2745 |
+
{
|
| 2746 |
+
"epoch": 0.002818009043842978,
|
| 2747 |
+
"grad_norm": 8.437786102294922,
|
| 2748 |
+
"learning_rate": 0.00019999963114969432,
|
| 2749 |
+
"loss": 1.1114,
|
| 2750 |
+
"step": 387
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 0.002825290720958851,
|
| 2754 |
+
"grad_norm": 12.213990211486816,
|
| 2755 |
+
"learning_rate": 0.00019999963114969432,
|
| 2756 |
+
"loss": 1.9927,
|
| 2757 |
+
"step": 388
|
| 2758 |
+
},
|
| 2759 |
+
{
|
| 2760 |
+
"epoch": 0.0028325723980747244,
|
| 2761 |
+
"grad_norm": 8.3518648147583,
|
| 2762 |
+
"learning_rate": 0.00019999963114969432,
|
| 2763 |
+
"loss": 1.3671,
|
| 2764 |
+
"step": 389
|
| 2765 |
+
},
|
| 2766 |
+
{
|
| 2767 |
+
"epoch": 0.002839854075190598,
|
| 2768 |
+
"grad_norm": 17.662803649902344,
|
| 2769 |
+
"learning_rate": 0.0001999996165977791,
|
| 2770 |
+
"loss": 2.1279,
|
| 2771 |
+
"step": 390
|
| 2772 |
+
},
|
| 2773 |
+
{
|
| 2774 |
+
"epoch": 0.0028471357523064712,
|
| 2775 |
+
"grad_norm": 8.939593315124512,
|
| 2776 |
+
"learning_rate": 0.0001999996165977791,
|
| 2777 |
+
"loss": 1.543,
|
| 2778 |
+
"step": 391
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 0.0028544174294223446,
|
| 2782 |
+
"grad_norm": 11.475914001464844,
|
| 2783 |
+
"learning_rate": 0.0001999996165977791,
|
| 2784 |
+
"loss": 1.6749,
|
| 2785 |
+
"step": 392
|
| 2786 |
+
},
|
| 2787 |
+
{
|
| 2788 |
+
"epoch": 0.002861699106538218,
|
| 2789 |
+
"grad_norm": 9.165367126464844,
|
| 2790 |
+
"learning_rate": 0.0001999996165977791,
|
| 2791 |
+
"loss": 1.7026,
|
| 2792 |
+
"step": 393
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 0.0028689807836540914,
|
| 2796 |
+
"grad_norm": 16.476228713989258,
|
| 2797 |
+
"learning_rate": 0.0001999996165977791,
|
| 2798 |
+
"loss": 2.8899,
|
| 2799 |
+
"step": 394
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 0.0028762624607699644,
|
| 2803 |
+
"grad_norm": 12.431415557861328,
|
| 2804 |
+
"learning_rate": 0.0001999996165977791,
|
| 2805 |
+
"loss": 2.3806,
|
| 2806 |
+
"step": 395
|
| 2807 |
+
},
|
| 2808 |
+
{
|
| 2809 |
+
"epoch": 0.0028835441378858378,
|
| 2810 |
+
"grad_norm": 14.651118278503418,
|
| 2811 |
+
"learning_rate": 0.00019999960204586387,
|
| 2812 |
+
"loss": 2.7375,
|
| 2813 |
+
"step": 396
|
| 2814 |
+
},
|
| 2815 |
+
{
|
| 2816 |
+
"epoch": 0.002890825815001711,
|
| 2817 |
+
"grad_norm": 11.111295700073242,
|
| 2818 |
+
"learning_rate": 0.00019999960204586387,
|
| 2819 |
+
"loss": 2.4693,
|
| 2820 |
+
"step": 397
|
| 2821 |
+
},
|
| 2822 |
+
{
|
| 2823 |
+
"epoch": 0.0028981074921175846,
|
| 2824 |
+
"grad_norm": 7.806118011474609,
|
| 2825 |
+
"learning_rate": 0.00019999960204586387,
|
| 2826 |
+
"loss": 1.7636,
|
| 2827 |
+
"step": 398
|
| 2828 |
+
},
|
| 2829 |
+
{
|
| 2830 |
+
"epoch": 0.002905389169233458,
|
| 2831 |
+
"grad_norm": 16.599353790283203,
|
| 2832 |
+
"learning_rate": 0.00019999960204586387,
|
| 2833 |
+
"loss": 2.8746,
|
| 2834 |
+
"step": 399
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 0.0029126708463493313,
|
| 2838 |
+
"grad_norm": 10.182330131530762,
|
| 2839 |
+
"learning_rate": 0.00019999960204586387,
|
| 2840 |
+
"loss": 1.6883,
|
| 2841 |
+
"step": 400
|
| 2842 |
+
},
|
| 2843 |
+
{
|
| 2844 |
+
"epoch": 0.0029126708463493313,
|
| 2845 |
+
"eval_loss": 1.8433881998062134,
|
| 2846 |
+
"eval_runtime": 36.2821,
|
| 2847 |
+
"eval_samples_per_second": 5.705,
|
| 2848 |
+
"eval_steps_per_second": 1.902,
|
| 2849 |
+
"step": 400
|
| 2850 |
}
|
| 2851 |
],
|
| 2852 |
"logging_steps": 1,
|
|
|
|
| 2861 |
"early_stopping_threshold": 0.0
|
| 2862 |
},
|
| 2863 |
"attributes": {
|
| 2864 |
+
"early_stopping_patience_counter": 3
|
| 2865 |
}
|
| 2866 |
},
|
| 2867 |
"TrainerControl": {
|
|
|
|
| 2870 |
"should_evaluate": false,
|
| 2871 |
"should_log": false,
|
| 2872 |
"should_save": true,
|
| 2873 |
+
"should_training_stop": true
|
| 2874 |
},
|
| 2875 |
"attributes": {}
|
| 2876 |
}
|
| 2877 |
},
|
| 2878 |
+
"total_flos": 2.65220203216896e+16,
|
| 2879 |
"train_batch_size": 3,
|
| 2880 |
"trial_name": null,
|
| 2881 |
"trial_params": null
|