Uploaded checkpoint-4000
Browse files- model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +711 -3
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2692969128
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3201017b6d299d8c2cb5eff8dfb87a841857ac5b48a904337a63577b57e72464
|
| 3 |
size 2692969128
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5386075202
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0bc91b5189b6a58b603e0148037300250d28c3b0c30d72a7e486062f8ab49769
|
| 3 |
size 5386075202
|
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 sha256:1c92a95a97d689d636b085d406167a1d143dce26fb83ee64d21cf4b37a120302
|
| 3 |
size 14244
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2f80b0441e18382140898e5947e4bf00161c8985bfd13094069daa8dad861cc8
|
| 3 |
size 1064
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.01655612699687481,
|
| 3 |
"best_model_checkpoint": "runs/deepseek_20240423-162824/checkpoint-3000",
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 1000,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -2131,6 +2131,714 @@
|
|
| 2131 |
"eval_samples_per_second": 18.745,
|
| 2132 |
"eval_steps_per_second": 18.745,
|
| 2133 |
"step": 3000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2134 |
}
|
| 2135 |
],
|
| 2136 |
"logging_steps": 10,
|
|
@@ -2138,7 +2846,7 @@
|
|
| 2138 |
"num_input_tokens_seen": 0,
|
| 2139 |
"num_train_epochs": 2,
|
| 2140 |
"save_steps": 1000,
|
| 2141 |
-
"total_flos":
|
| 2142 |
"train_batch_size": 1,
|
| 2143 |
"trial_name": null,
|
| 2144 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.01655612699687481,
|
| 3 |
"best_model_checkpoint": "runs/deepseek_20240423-162824/checkpoint-3000",
|
| 4 |
+
"epoch": 1.2532309861361322,
|
| 5 |
"eval_steps": 1000,
|
| 6 |
+
"global_step": 4000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2131 |
"eval_samples_per_second": 18.745,
|
| 2132 |
"eval_steps_per_second": 18.745,
|
| 2133 |
"step": 3000
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 0.94,
|
| 2137 |
+
"grad_norm": 10.6875,
|
| 2138 |
+
"learning_rate": 8.844444444444445e-06,
|
| 2139 |
+
"loss": 0.137,
|
| 2140 |
+
"step": 3010
|
| 2141 |
+
},
|
| 2142 |
+
{
|
| 2143 |
+
"epoch": 0.95,
|
| 2144 |
+
"grad_norm": 0.00127410888671875,
|
| 2145 |
+
"learning_rate": 8.8e-06,
|
| 2146 |
+
"loss": 0.0456,
|
| 2147 |
+
"step": 3020
|
| 2148 |
+
},
|
| 2149 |
+
{
|
| 2150 |
+
"epoch": 0.95,
|
| 2151 |
+
"grad_norm": 0.003204345703125,
|
| 2152 |
+
"learning_rate": 8.755555555555556e-06,
|
| 2153 |
+
"loss": 0.0133,
|
| 2154 |
+
"step": 3030
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 0.95,
|
| 2158 |
+
"grad_norm": 8.625,
|
| 2159 |
+
"learning_rate": 8.711111111111111e-06,
|
| 2160 |
+
"loss": 0.1079,
|
| 2161 |
+
"step": 3040
|
| 2162 |
+
},
|
| 2163 |
+
{
|
| 2164 |
+
"epoch": 0.96,
|
| 2165 |
+
"grad_norm": 0.004791259765625,
|
| 2166 |
+
"learning_rate": 8.666666666666668e-06,
|
| 2167 |
+
"loss": 0.0327,
|
| 2168 |
+
"step": 3050
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 0.96,
|
| 2172 |
+
"grad_norm": 0.004150390625,
|
| 2173 |
+
"learning_rate": 8.622222222222223e-06,
|
| 2174 |
+
"loss": 0.0525,
|
| 2175 |
+
"step": 3060
|
| 2176 |
+
},
|
| 2177 |
+
{
|
| 2178 |
+
"epoch": 0.96,
|
| 2179 |
+
"grad_norm": 8.9375,
|
| 2180 |
+
"learning_rate": 8.577777777777778e-06,
|
| 2181 |
+
"loss": 0.0584,
|
| 2182 |
+
"step": 3070
|
| 2183 |
+
},
|
| 2184 |
+
{
|
| 2185 |
+
"epoch": 0.96,
|
| 2186 |
+
"grad_norm": 0.00131988525390625,
|
| 2187 |
+
"learning_rate": 8.533333333333335e-06,
|
| 2188 |
+
"loss": 0.0406,
|
| 2189 |
+
"step": 3080
|
| 2190 |
+
},
|
| 2191 |
+
{
|
| 2192 |
+
"epoch": 0.97,
|
| 2193 |
+
"grad_norm": 14.0,
|
| 2194 |
+
"learning_rate": 8.48888888888889e-06,
|
| 2195 |
+
"loss": 0.0195,
|
| 2196 |
+
"step": 3090
|
| 2197 |
+
},
|
| 2198 |
+
{
|
| 2199 |
+
"epoch": 0.97,
|
| 2200 |
+
"grad_norm": 0.001373291015625,
|
| 2201 |
+
"learning_rate": 8.444444444444446e-06,
|
| 2202 |
+
"loss": 0.0571,
|
| 2203 |
+
"step": 3100
|
| 2204 |
+
},
|
| 2205 |
+
{
|
| 2206 |
+
"epoch": 0.97,
|
| 2207 |
+
"grad_norm": 0.00469970703125,
|
| 2208 |
+
"learning_rate": 8.400000000000001e-06,
|
| 2209 |
+
"loss": 0.1144,
|
| 2210 |
+
"step": 3110
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 0.98,
|
| 2214 |
+
"grad_norm": 0.03759765625,
|
| 2215 |
+
"learning_rate": 8.355555555555556e-06,
|
| 2216 |
+
"loss": 0.0206,
|
| 2217 |
+
"step": 3120
|
| 2218 |
+
},
|
| 2219 |
+
{
|
| 2220 |
+
"epoch": 0.98,
|
| 2221 |
+
"grad_norm": 0.0027618408203125,
|
| 2222 |
+
"learning_rate": 8.311111111111111e-06,
|
| 2223 |
+
"loss": 0.0463,
|
| 2224 |
+
"step": 3130
|
| 2225 |
+
},
|
| 2226 |
+
{
|
| 2227 |
+
"epoch": 0.98,
|
| 2228 |
+
"grad_norm": 0.00156402587890625,
|
| 2229 |
+
"learning_rate": 8.266666666666667e-06,
|
| 2230 |
+
"loss": 0.079,
|
| 2231 |
+
"step": 3140
|
| 2232 |
+
},
|
| 2233 |
+
{
|
| 2234 |
+
"epoch": 0.99,
|
| 2235 |
+
"grad_norm": 12.875,
|
| 2236 |
+
"learning_rate": 8.222222222222222e-06,
|
| 2237 |
+
"loss": 0.0177,
|
| 2238 |
+
"step": 3150
|
| 2239 |
+
},
|
| 2240 |
+
{
|
| 2241 |
+
"epoch": 0.99,
|
| 2242 |
+
"grad_norm": 0.004547119140625,
|
| 2243 |
+
"learning_rate": 8.177777777777779e-06,
|
| 2244 |
+
"loss": 0.0287,
|
| 2245 |
+
"step": 3160
|
| 2246 |
+
},
|
| 2247 |
+
{
|
| 2248 |
+
"epoch": 0.99,
|
| 2249 |
+
"grad_norm": 0.00136566162109375,
|
| 2250 |
+
"learning_rate": 8.133333333333334e-06,
|
| 2251 |
+
"loss": 0.0063,
|
| 2252 |
+
"step": 3170
|
| 2253 |
+
},
|
| 2254 |
+
{
|
| 2255 |
+
"epoch": 1.0,
|
| 2256 |
+
"grad_norm": 0.00140380859375,
|
| 2257 |
+
"learning_rate": 8.08888888888889e-06,
|
| 2258 |
+
"loss": 0.0575,
|
| 2259 |
+
"step": 3180
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 1.0,
|
| 2263 |
+
"grad_norm": 0.00157928466796875,
|
| 2264 |
+
"learning_rate": 8.044444444444444e-06,
|
| 2265 |
+
"loss": 0.0281,
|
| 2266 |
+
"step": 3190
|
| 2267 |
+
},
|
| 2268 |
+
{
|
| 2269 |
+
"epoch": 1.0,
|
| 2270 |
+
"grad_norm": 4.34375,
|
| 2271 |
+
"learning_rate": 8.000000000000001e-06,
|
| 2272 |
+
"loss": 0.0165,
|
| 2273 |
+
"step": 3200
|
| 2274 |
+
},
|
| 2275 |
+
{
|
| 2276 |
+
"epoch": 1.01,
|
| 2277 |
+
"grad_norm": 1.2109375,
|
| 2278 |
+
"learning_rate": 7.955555555555557e-06,
|
| 2279 |
+
"loss": 0.0136,
|
| 2280 |
+
"step": 3210
|
| 2281 |
+
},
|
| 2282 |
+
{
|
| 2283 |
+
"epoch": 1.01,
|
| 2284 |
+
"grad_norm": 28.875,
|
| 2285 |
+
"learning_rate": 7.911111111111112e-06,
|
| 2286 |
+
"loss": 0.0679,
|
| 2287 |
+
"step": 3220
|
| 2288 |
+
},
|
| 2289 |
+
{
|
| 2290 |
+
"epoch": 1.01,
|
| 2291 |
+
"grad_norm": 0.001251220703125,
|
| 2292 |
+
"learning_rate": 7.866666666666667e-06,
|
| 2293 |
+
"loss": 0.0122,
|
| 2294 |
+
"step": 3230
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 1.02,
|
| 2298 |
+
"grad_norm": 3.703125,
|
| 2299 |
+
"learning_rate": 7.822222222222224e-06,
|
| 2300 |
+
"loss": 0.0085,
|
| 2301 |
+
"step": 3240
|
| 2302 |
+
},
|
| 2303 |
+
{
|
| 2304 |
+
"epoch": 1.02,
|
| 2305 |
+
"grad_norm": 0.00482177734375,
|
| 2306 |
+
"learning_rate": 7.77777777777778e-06,
|
| 2307 |
+
"loss": 0.0195,
|
| 2308 |
+
"step": 3250
|
| 2309 |
+
},
|
| 2310 |
+
{
|
| 2311 |
+
"epoch": 1.02,
|
| 2312 |
+
"grad_norm": 0.06396484375,
|
| 2313 |
+
"learning_rate": 7.733333333333334e-06,
|
| 2314 |
+
"loss": 0.0234,
|
| 2315 |
+
"step": 3260
|
| 2316 |
+
},
|
| 2317 |
+
{
|
| 2318 |
+
"epoch": 1.02,
|
| 2319 |
+
"grad_norm": 4.71875,
|
| 2320 |
+
"learning_rate": 7.68888888888889e-06,
|
| 2321 |
+
"loss": 0.0451,
|
| 2322 |
+
"step": 3270
|
| 2323 |
+
},
|
| 2324 |
+
{
|
| 2325 |
+
"epoch": 1.03,
|
| 2326 |
+
"grad_norm": 0.00148773193359375,
|
| 2327 |
+
"learning_rate": 7.644444444444445e-06,
|
| 2328 |
+
"loss": 0.0098,
|
| 2329 |
+
"step": 3280
|
| 2330 |
+
},
|
| 2331 |
+
{
|
| 2332 |
+
"epoch": 1.03,
|
| 2333 |
+
"grad_norm": 0.002532958984375,
|
| 2334 |
+
"learning_rate": 7.600000000000001e-06,
|
| 2335 |
+
"loss": 0.0034,
|
| 2336 |
+
"step": 3290
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 1.03,
|
| 2340 |
+
"grad_norm": 0.0027923583984375,
|
| 2341 |
+
"learning_rate": 7.555555555555556e-06,
|
| 2342 |
+
"loss": 0.01,
|
| 2343 |
+
"step": 3300
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 1.04,
|
| 2347 |
+
"grad_norm": 3.5,
|
| 2348 |
+
"learning_rate": 7.511111111111111e-06,
|
| 2349 |
+
"loss": 0.0269,
|
| 2350 |
+
"step": 3310
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 1.04,
|
| 2354 |
+
"grad_norm": 0.00124359130859375,
|
| 2355 |
+
"learning_rate": 7.4666666666666675e-06,
|
| 2356 |
+
"loss": 0.018,
|
| 2357 |
+
"step": 3320
|
| 2358 |
+
},
|
| 2359 |
+
{
|
| 2360 |
+
"epoch": 1.04,
|
| 2361 |
+
"grad_norm": 8.75,
|
| 2362 |
+
"learning_rate": 7.422222222222223e-06,
|
| 2363 |
+
"loss": 0.1035,
|
| 2364 |
+
"step": 3330
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 1.05,
|
| 2368 |
+
"grad_norm": 13.875,
|
| 2369 |
+
"learning_rate": 7.377777777777778e-06,
|
| 2370 |
+
"loss": 0.0368,
|
| 2371 |
+
"step": 3340
|
| 2372 |
+
},
|
| 2373 |
+
{
|
| 2374 |
+
"epoch": 1.05,
|
| 2375 |
+
"grad_norm": 0.004608154296875,
|
| 2376 |
+
"learning_rate": 7.333333333333333e-06,
|
| 2377 |
+
"loss": 0.0243,
|
| 2378 |
+
"step": 3350
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 1.05,
|
| 2382 |
+
"grad_norm": 0.00146484375,
|
| 2383 |
+
"learning_rate": 7.28888888888889e-06,
|
| 2384 |
+
"loss": 0.0177,
|
| 2385 |
+
"step": 3360
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 1.06,
|
| 2389 |
+
"grad_norm": 0.0014190673828125,
|
| 2390 |
+
"learning_rate": 7.244444444444445e-06,
|
| 2391 |
+
"loss": 0.0426,
|
| 2392 |
+
"step": 3370
|
| 2393 |
+
},
|
| 2394 |
+
{
|
| 2395 |
+
"epoch": 1.06,
|
| 2396 |
+
"grad_norm": 0.004913330078125,
|
| 2397 |
+
"learning_rate": 7.2000000000000005e-06,
|
| 2398 |
+
"loss": 0.0506,
|
| 2399 |
+
"step": 3380
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 1.06,
|
| 2403 |
+
"grad_norm": 9.125,
|
| 2404 |
+
"learning_rate": 7.155555555555556e-06,
|
| 2405 |
+
"loss": 0.0113,
|
| 2406 |
+
"step": 3390
|
| 2407 |
+
},
|
| 2408 |
+
{
|
| 2409 |
+
"epoch": 1.07,
|
| 2410 |
+
"grad_norm": 0.0015716552734375,
|
| 2411 |
+
"learning_rate": 7.111111111111112e-06,
|
| 2412 |
+
"loss": 0.0075,
|
| 2413 |
+
"step": 3400
|
| 2414 |
+
},
|
| 2415 |
+
{
|
| 2416 |
+
"epoch": 1.07,
|
| 2417 |
+
"grad_norm": 9.9375,
|
| 2418 |
+
"learning_rate": 7.066666666666667e-06,
|
| 2419 |
+
"loss": 0.0188,
|
| 2420 |
+
"step": 3410
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 1.07,
|
| 2424 |
+
"grad_norm": 0.001312255859375,
|
| 2425 |
+
"learning_rate": 7.022222222222222e-06,
|
| 2426 |
+
"loss": 0.0017,
|
| 2427 |
+
"step": 3420
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 1.07,
|
| 2431 |
+
"grad_norm": 0.001251220703125,
|
| 2432 |
+
"learning_rate": 6.977777777777779e-06,
|
| 2433 |
+
"loss": 0.0339,
|
| 2434 |
+
"step": 3430
|
| 2435 |
+
},
|
| 2436 |
+
{
|
| 2437 |
+
"epoch": 1.08,
|
| 2438 |
+
"grad_norm": 0.00262451171875,
|
| 2439 |
+
"learning_rate": 6.9333333333333344e-06,
|
| 2440 |
+
"loss": 0.0326,
|
| 2441 |
+
"step": 3440
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 1.08,
|
| 2445 |
+
"grad_norm": 0.004547119140625,
|
| 2446 |
+
"learning_rate": 6.88888888888889e-06,
|
| 2447 |
+
"loss": 0.0145,
|
| 2448 |
+
"step": 3450
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 1.08,
|
| 2452 |
+
"grad_norm": 0.00107574462890625,
|
| 2453 |
+
"learning_rate": 6.844444444444445e-06,
|
| 2454 |
+
"loss": 0.0126,
|
| 2455 |
+
"step": 3460
|
| 2456 |
+
},
|
| 2457 |
+
{
|
| 2458 |
+
"epoch": 1.09,
|
| 2459 |
+
"grad_norm": 0.875,
|
| 2460 |
+
"learning_rate": 6.800000000000001e-06,
|
| 2461 |
+
"loss": 0.0141,
|
| 2462 |
+
"step": 3470
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 1.09,
|
| 2466 |
+
"grad_norm": 7.59375,
|
| 2467 |
+
"learning_rate": 6.755555555555556e-06,
|
| 2468 |
+
"loss": 0.0301,
|
| 2469 |
+
"step": 3480
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 1.09,
|
| 2473 |
+
"grad_norm": 7.75,
|
| 2474 |
+
"learning_rate": 6.711111111111111e-06,
|
| 2475 |
+
"loss": 0.0432,
|
| 2476 |
+
"step": 3490
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 1.1,
|
| 2480 |
+
"grad_norm": 0.0010528564453125,
|
| 2481 |
+
"learning_rate": 6.666666666666667e-06,
|
| 2482 |
+
"loss": 0.0335,
|
| 2483 |
+
"step": 3500
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 1.1,
|
| 2487 |
+
"grad_norm": 0.0048828125,
|
| 2488 |
+
"learning_rate": 6.6222222222222236e-06,
|
| 2489 |
+
"loss": 0.0616,
|
| 2490 |
+
"step": 3510
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 1.1,
|
| 2494 |
+
"grad_norm": 0.001312255859375,
|
| 2495 |
+
"learning_rate": 6.577777777777779e-06,
|
| 2496 |
+
"loss": 0.0144,
|
| 2497 |
+
"step": 3520
|
| 2498 |
+
},
|
| 2499 |
+
{
|
| 2500 |
+
"epoch": 1.11,
|
| 2501 |
+
"grad_norm": 2.5,
|
| 2502 |
+
"learning_rate": 6.533333333333334e-06,
|
| 2503 |
+
"loss": 0.0415,
|
| 2504 |
+
"step": 3530
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 1.11,
|
| 2508 |
+
"grad_norm": 0.004730224609375,
|
| 2509 |
+
"learning_rate": 6.488888888888889e-06,
|
| 2510 |
+
"loss": 0.0315,
|
| 2511 |
+
"step": 3540
|
| 2512 |
+
},
|
| 2513 |
+
{
|
| 2514 |
+
"epoch": 1.11,
|
| 2515 |
+
"grad_norm": 0.00128173828125,
|
| 2516 |
+
"learning_rate": 6.444444444444445e-06,
|
| 2517 |
+
"loss": 0.0051,
|
| 2518 |
+
"step": 3550
|
| 2519 |
+
},
|
| 2520 |
+
{
|
| 2521 |
+
"epoch": 1.12,
|
| 2522 |
+
"grad_norm": 0.00115966796875,
|
| 2523 |
+
"learning_rate": 6.4000000000000006e-06,
|
| 2524 |
+
"loss": 0.0411,
|
| 2525 |
+
"step": 3560
|
| 2526 |
+
},
|
| 2527 |
+
{
|
| 2528 |
+
"epoch": 1.12,
|
| 2529 |
+
"grad_norm": 0.0010528564453125,
|
| 2530 |
+
"learning_rate": 6.355555555555556e-06,
|
| 2531 |
+
"loss": 0.0271,
|
| 2532 |
+
"step": 3570
|
| 2533 |
+
},
|
| 2534 |
+
{
|
| 2535 |
+
"epoch": 1.12,
|
| 2536 |
+
"grad_norm": 0.00482177734375,
|
| 2537 |
+
"learning_rate": 6.311111111111111e-06,
|
| 2538 |
+
"loss": 0.0172,
|
| 2539 |
+
"step": 3580
|
| 2540 |
+
},
|
| 2541 |
+
{
|
| 2542 |
+
"epoch": 1.12,
|
| 2543 |
+
"grad_norm": 0.0027008056640625,
|
| 2544 |
+
"learning_rate": 6.266666666666668e-06,
|
| 2545 |
+
"loss": 0.0062,
|
| 2546 |
+
"step": 3590
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 1.13,
|
| 2550 |
+
"grad_norm": 6.84375,
|
| 2551 |
+
"learning_rate": 6.222222222222223e-06,
|
| 2552 |
+
"loss": 0.0234,
|
| 2553 |
+
"step": 3600
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 1.13,
|
| 2557 |
+
"grad_norm": 0.0023345947265625,
|
| 2558 |
+
"learning_rate": 6.177777777777778e-06,
|
| 2559 |
+
"loss": 0.0,
|
| 2560 |
+
"step": 3610
|
| 2561 |
+
},
|
| 2562 |
+
{
|
| 2563 |
+
"epoch": 1.13,
|
| 2564 |
+
"grad_norm": 0.00115966796875,
|
| 2565 |
+
"learning_rate": 6.133333333333334e-06,
|
| 2566 |
+
"loss": 0.0194,
|
| 2567 |
+
"step": 3620
|
| 2568 |
+
},
|
| 2569 |
+
{
|
| 2570 |
+
"epoch": 1.14,
|
| 2571 |
+
"grad_norm": 0.1376953125,
|
| 2572 |
+
"learning_rate": 6.08888888888889e-06,
|
| 2573 |
+
"loss": 0.0088,
|
| 2574 |
+
"step": 3630
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 1.14,
|
| 2578 |
+
"grad_norm": 0.203125,
|
| 2579 |
+
"learning_rate": 6.044444444444445e-06,
|
| 2580 |
+
"loss": 0.0339,
|
| 2581 |
+
"step": 3640
|
| 2582 |
+
},
|
| 2583 |
+
{
|
| 2584 |
+
"epoch": 1.14,
|
| 2585 |
+
"grad_norm": 15.25,
|
| 2586 |
+
"learning_rate": 6e-06,
|
| 2587 |
+
"loss": 0.0362,
|
| 2588 |
+
"step": 3650
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 1.15,
|
| 2592 |
+
"grad_norm": 4.375,
|
| 2593 |
+
"learning_rate": 5.955555555555555e-06,
|
| 2594 |
+
"loss": 0.0335,
|
| 2595 |
+
"step": 3660
|
| 2596 |
+
},
|
| 2597 |
+
{
|
| 2598 |
+
"epoch": 1.15,
|
| 2599 |
+
"grad_norm": 3.5,
|
| 2600 |
+
"learning_rate": 5.911111111111112e-06,
|
| 2601 |
+
"loss": 0.0254,
|
| 2602 |
+
"step": 3670
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 1.15,
|
| 2606 |
+
"grad_norm": 0.000949859619140625,
|
| 2607 |
+
"learning_rate": 5.8666666666666675e-06,
|
| 2608 |
+
"loss": 0.0173,
|
| 2609 |
+
"step": 3680
|
| 2610 |
+
},
|
| 2611 |
+
{
|
| 2612 |
+
"epoch": 1.16,
|
| 2613 |
+
"grad_norm": 5.125,
|
| 2614 |
+
"learning_rate": 5.822222222222223e-06,
|
| 2615 |
+
"loss": 0.0677,
|
| 2616 |
+
"step": 3690
|
| 2617 |
+
},
|
| 2618 |
+
{
|
| 2619 |
+
"epoch": 1.16,
|
| 2620 |
+
"grad_norm": 9.75,
|
| 2621 |
+
"learning_rate": 5.777777777777778e-06,
|
| 2622 |
+
"loss": 0.0162,
|
| 2623 |
+
"step": 3700
|
| 2624 |
+
},
|
| 2625 |
+
{
|
| 2626 |
+
"epoch": 1.16,
|
| 2627 |
+
"grad_norm": 0.002288818359375,
|
| 2628 |
+
"learning_rate": 5.733333333333334e-06,
|
| 2629 |
+
"loss": 0.0334,
|
| 2630 |
+
"step": 3710
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"epoch": 1.17,
|
| 2634 |
+
"grad_norm": 0.0230712890625,
|
| 2635 |
+
"learning_rate": 5.688888888888889e-06,
|
| 2636 |
+
"loss": 0.0673,
|
| 2637 |
+
"step": 3720
|
| 2638 |
+
},
|
| 2639 |
+
{
|
| 2640 |
+
"epoch": 1.17,
|
| 2641 |
+
"grad_norm": 0.0010986328125,
|
| 2642 |
+
"learning_rate": 5.6444444444444445e-06,
|
| 2643 |
+
"loss": 0.0189,
|
| 2644 |
+
"step": 3730
|
| 2645 |
+
},
|
| 2646 |
+
{
|
| 2647 |
+
"epoch": 1.17,
|
| 2648 |
+
"grad_norm": 0.00095367431640625,
|
| 2649 |
+
"learning_rate": 5.600000000000001e-06,
|
| 2650 |
+
"loss": 0.0493,
|
| 2651 |
+
"step": 3740
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 1.17,
|
| 2655 |
+
"grad_norm": 0.0093994140625,
|
| 2656 |
+
"learning_rate": 5.555555555555557e-06,
|
| 2657 |
+
"loss": 0.032,
|
| 2658 |
+
"step": 3750
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 1.18,
|
| 2662 |
+
"grad_norm": 0.00119781494140625,
|
| 2663 |
+
"learning_rate": 5.511111111111112e-06,
|
| 2664 |
+
"loss": 0.0201,
|
| 2665 |
+
"step": 3760
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 1.18,
|
| 2669 |
+
"grad_norm": 0.00225830078125,
|
| 2670 |
+
"learning_rate": 5.466666666666667e-06,
|
| 2671 |
+
"loss": 0.0928,
|
| 2672 |
+
"step": 3770
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 1.18,
|
| 2676 |
+
"grad_norm": 0.0045166015625,
|
| 2677 |
+
"learning_rate": 5.422222222222223e-06,
|
| 2678 |
+
"loss": 0.0048,
|
| 2679 |
+
"step": 3780
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 1.19,
|
| 2683 |
+
"grad_norm": 0.1513671875,
|
| 2684 |
+
"learning_rate": 5.3777777777777784e-06,
|
| 2685 |
+
"loss": 0.023,
|
| 2686 |
+
"step": 3790
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 1.19,
|
| 2690 |
+
"grad_norm": 2.875,
|
| 2691 |
+
"learning_rate": 5.333333333333334e-06,
|
| 2692 |
+
"loss": 0.0134,
|
| 2693 |
+
"step": 3800
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 1.19,
|
| 2697 |
+
"grad_norm": 0.00146484375,
|
| 2698 |
+
"learning_rate": 5.288888888888889e-06,
|
| 2699 |
+
"loss": 0.0063,
|
| 2700 |
+
"step": 3810
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 1.2,
|
| 2704 |
+
"grad_norm": 0.001007080078125,
|
| 2705 |
+
"learning_rate": 5.244444444444445e-06,
|
| 2706 |
+
"loss": 0.0241,
|
| 2707 |
+
"step": 3820
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 1.2,
|
| 2711 |
+
"grad_norm": 0.01483154296875,
|
| 2712 |
+
"learning_rate": 5.2e-06,
|
| 2713 |
+
"loss": 0.0124,
|
| 2714 |
+
"step": 3830
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 1.2,
|
| 2718 |
+
"grad_norm": 0.00433349609375,
|
| 2719 |
+
"learning_rate": 5.155555555555556e-06,
|
| 2720 |
+
"loss": 0.0231,
|
| 2721 |
+
"step": 3840
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 1.21,
|
| 2725 |
+
"grad_norm": 0.00201416015625,
|
| 2726 |
+
"learning_rate": 5.1111111111111115e-06,
|
| 2727 |
+
"loss": 0.0001,
|
| 2728 |
+
"step": 3850
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 1.21,
|
| 2732 |
+
"grad_norm": 4.375,
|
| 2733 |
+
"learning_rate": 5.0666666666666676e-06,
|
| 2734 |
+
"loss": 0.0294,
|
| 2735 |
+
"step": 3860
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 1.21,
|
| 2739 |
+
"grad_norm": 0.004547119140625,
|
| 2740 |
+
"learning_rate": 5.022222222222223e-06,
|
| 2741 |
+
"loss": 0.0379,
|
| 2742 |
+
"step": 3870
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 1.22,
|
| 2746 |
+
"grad_norm": 0.0010223388671875,
|
| 2747 |
+
"learning_rate": 4.977777777777778e-06,
|
| 2748 |
+
"loss": 0.0284,
|
| 2749 |
+
"step": 3880
|
| 2750 |
+
},
|
| 2751 |
+
{
|
| 2752 |
+
"epoch": 1.22,
|
| 2753 |
+
"grad_norm": 4.8125,
|
| 2754 |
+
"learning_rate": 4.933333333333334e-06,
|
| 2755 |
+
"loss": 0.0131,
|
| 2756 |
+
"step": 3890
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 1.22,
|
| 2760 |
+
"grad_norm": 3.9375,
|
| 2761 |
+
"learning_rate": 4.888888888888889e-06,
|
| 2762 |
+
"loss": 0.0527,
|
| 2763 |
+
"step": 3900
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 1.23,
|
| 2767 |
+
"grad_norm": 0.0010986328125,
|
| 2768 |
+
"learning_rate": 4.8444444444444446e-06,
|
| 2769 |
+
"loss": 0.0002,
|
| 2770 |
+
"step": 3910
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 1.23,
|
| 2774 |
+
"grad_norm": 0.0035247802734375,
|
| 2775 |
+
"learning_rate": 4.800000000000001e-06,
|
| 2776 |
+
"loss": 0.0455,
|
| 2777 |
+
"step": 3920
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 1.23,
|
| 2781 |
+
"grad_norm": 8.125,
|
| 2782 |
+
"learning_rate": 4.755555555555556e-06,
|
| 2783 |
+
"loss": 0.0104,
|
| 2784 |
+
"step": 3930
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 1.23,
|
| 2788 |
+
"grad_norm": 0.00156402587890625,
|
| 2789 |
+
"learning_rate": 4.711111111111111e-06,
|
| 2790 |
+
"loss": 0.0,
|
| 2791 |
+
"step": 3940
|
| 2792 |
+
},
|
| 2793 |
+
{
|
| 2794 |
+
"epoch": 1.24,
|
| 2795 |
+
"grad_norm": 0.0026397705078125,
|
| 2796 |
+
"learning_rate": 4.666666666666667e-06,
|
| 2797 |
+
"loss": 0.0191,
|
| 2798 |
+
"step": 3950
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 1.24,
|
| 2802 |
+
"grad_norm": 0.0037689208984375,
|
| 2803 |
+
"learning_rate": 4.622222222222222e-06,
|
| 2804 |
+
"loss": 0.0288,
|
| 2805 |
+
"step": 3960
|
| 2806 |
+
},
|
| 2807 |
+
{
|
| 2808 |
+
"epoch": 1.24,
|
| 2809 |
+
"grad_norm": 0.00135040283203125,
|
| 2810 |
+
"learning_rate": 4.5777777777777785e-06,
|
| 2811 |
+
"loss": 0.0519,
|
| 2812 |
+
"step": 3970
|
| 2813 |
+
},
|
| 2814 |
+
{
|
| 2815 |
+
"epoch": 1.25,
|
| 2816 |
+
"grad_norm": 0.00106048583984375,
|
| 2817 |
+
"learning_rate": 4.533333333333334e-06,
|
| 2818 |
+
"loss": 0.0175,
|
| 2819 |
+
"step": 3980
|
| 2820 |
+
},
|
| 2821 |
+
{
|
| 2822 |
+
"epoch": 1.25,
|
| 2823 |
+
"grad_norm": 7.0,
|
| 2824 |
+
"learning_rate": 4.488888888888889e-06,
|
| 2825 |
+
"loss": 0.0066,
|
| 2826 |
+
"step": 3990
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"epoch": 1.25,
|
| 2830 |
+
"grad_norm": 0.0308837890625,
|
| 2831 |
+
"learning_rate": 4.444444444444444e-06,
|
| 2832 |
+
"loss": 0.0253,
|
| 2833 |
+
"step": 4000
|
| 2834 |
+
},
|
| 2835 |
+
{
|
| 2836 |
+
"epoch": 1.25,
|
| 2837 |
+
"eval_loss": 0.019900379702448845,
|
| 2838 |
+
"eval_runtime": 53.4572,
|
| 2839 |
+
"eval_samples_per_second": 18.707,
|
| 2840 |
+
"eval_steps_per_second": 18.707,
|
| 2841 |
+
"step": 4000
|
| 2842 |
}
|
| 2843 |
],
|
| 2844 |
"logging_steps": 10,
|
|
|
|
| 2846 |
"num_input_tokens_seen": 0,
|
| 2847 |
"num_train_epochs": 2,
|
| 2848 |
"save_steps": 1000,
|
| 2849 |
+
"total_flos": 6.2934793322496e+16,
|
| 2850 |
"train_batch_size": 1,
|
| 2851 |
"trial_name": null,
|
| 2852 |
"trial_params": null
|