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 2836579040
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:34492fb18964a4396853a5dd6bd8c9f2527750844e4ca0b7edebd4fd8fd907bf
|
3 |
size 2836579040
|
optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5673376169
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:82a35ac1d4172f6fc7a72acb2fe839f1b7ed06c36b809b261d5e6781653067c6
|
3 |
size 5673376169
|
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 sha256:9de204c0067b7b80c3b397a5621bcbaed6c4c0e59bfc1ac45b0bfde553b2fcfa
|
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:c5c3d30b62ee42789c2e49d34a8f0fd414c5b2699878096091f51926dfd0a5c5
|
3 |
size 1064
|
trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
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": 15.5,
|
2132 |
"eval_steps_per_second": 15.5,
|
2133 |
"step": 3000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2134 |
}
|
2135 |
],
|
2136 |
"logging_steps": 10,
|
@@ -2138,7 +2846,7 @@
|
|
2138 |
"num_input_tokens_seen": 0,
|
2139 |
"num_train_epochs": 1,
|
2140 |
"save_steps": 1000,
|
2141 |
-
"total_flos":
|
2142 |
"train_batch_size": 1,
|
2143 |
"trial_name": null,
|
2144 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.2,
|
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": 15.5,
|
2132 |
"eval_steps_per_second": 15.5,
|
2133 |
"step": 3000
|
2134 |
+
},
|
2135 |
+
{
|
2136 |
+
"epoch": 0.15,
|
2137 |
+
"grad_norm": 5.84375,
|
2138 |
+
"learning_rate": 4.864444444444445e-07,
|
2139 |
+
"loss": 0.0486,
|
2140 |
+
"step": 3010
|
2141 |
+
},
|
2142 |
+
{
|
2143 |
+
"epoch": 0.15,
|
2144 |
+
"grad_norm": 6.34375,
|
2145 |
+
"learning_rate": 4.84e-07,
|
2146 |
+
"loss": 0.0504,
|
2147 |
+
"step": 3020
|
2148 |
+
},
|
2149 |
+
{
|
2150 |
+
"epoch": 0.15,
|
2151 |
+
"grad_norm": 6.15625,
|
2152 |
+
"learning_rate": 4.815555555555556e-07,
|
2153 |
+
"loss": 0.0576,
|
2154 |
+
"step": 3030
|
2155 |
+
},
|
2156 |
+
{
|
2157 |
+
"epoch": 0.15,
|
2158 |
+
"grad_norm": 5.4375,
|
2159 |
+
"learning_rate": 4.791111111111111e-07,
|
2160 |
+
"loss": 0.0585,
|
2161 |
+
"step": 3040
|
2162 |
+
},
|
2163 |
+
{
|
2164 |
+
"epoch": 0.15,
|
2165 |
+
"grad_norm": 5.25,
|
2166 |
+
"learning_rate": 4.766666666666667e-07,
|
2167 |
+
"loss": 0.0557,
|
2168 |
+
"step": 3050
|
2169 |
+
},
|
2170 |
+
{
|
2171 |
+
"epoch": 0.15,
|
2172 |
+
"grad_norm": 6.84375,
|
2173 |
+
"learning_rate": 4.7422222222222225e-07,
|
2174 |
+
"loss": 0.0509,
|
2175 |
+
"step": 3060
|
2176 |
+
},
|
2177 |
+
{
|
2178 |
+
"epoch": 0.15,
|
2179 |
+
"grad_norm": 8.6875,
|
2180 |
+
"learning_rate": 4.717777777777778e-07,
|
2181 |
+
"loss": 0.0596,
|
2182 |
+
"step": 3070
|
2183 |
+
},
|
2184 |
+
{
|
2185 |
+
"epoch": 0.15,
|
2186 |
+
"grad_norm": 8.1875,
|
2187 |
+
"learning_rate": 4.693333333333334e-07,
|
2188 |
+
"loss": 0.0532,
|
2189 |
+
"step": 3080
|
2190 |
+
},
|
2191 |
+
{
|
2192 |
+
"epoch": 0.15,
|
2193 |
+
"grad_norm": 5.40625,
|
2194 |
+
"learning_rate": 4.6688888888888893e-07,
|
2195 |
+
"loss": 0.0546,
|
2196 |
+
"step": 3090
|
2197 |
+
},
|
2198 |
+
{
|
2199 |
+
"epoch": 0.15,
|
2200 |
+
"grad_norm": 8.5,
|
2201 |
+
"learning_rate": 4.6444444444444446e-07,
|
2202 |
+
"loss": 0.0614,
|
2203 |
+
"step": 3100
|
2204 |
+
},
|
2205 |
+
{
|
2206 |
+
"epoch": 0.16,
|
2207 |
+
"grad_norm": 5.375,
|
2208 |
+
"learning_rate": 4.62e-07,
|
2209 |
+
"loss": 0.0513,
|
2210 |
+
"step": 3110
|
2211 |
+
},
|
2212 |
+
{
|
2213 |
+
"epoch": 0.16,
|
2214 |
+
"grad_norm": 6.71875,
|
2215 |
+
"learning_rate": 4.595555555555556e-07,
|
2216 |
+
"loss": 0.0542,
|
2217 |
+
"step": 3120
|
2218 |
+
},
|
2219 |
+
{
|
2220 |
+
"epoch": 0.16,
|
2221 |
+
"grad_norm": 5.21875,
|
2222 |
+
"learning_rate": 4.5711111111111114e-07,
|
2223 |
+
"loss": 0.0513,
|
2224 |
+
"step": 3130
|
2225 |
+
},
|
2226 |
+
{
|
2227 |
+
"epoch": 0.16,
|
2228 |
+
"grad_norm": 7.6875,
|
2229 |
+
"learning_rate": 4.5466666666666666e-07,
|
2230 |
+
"loss": 0.0589,
|
2231 |
+
"step": 3140
|
2232 |
+
},
|
2233 |
+
{
|
2234 |
+
"epoch": 0.16,
|
2235 |
+
"grad_norm": 6.09375,
|
2236 |
+
"learning_rate": 4.5222222222222224e-07,
|
2237 |
+
"loss": 0.0515,
|
2238 |
+
"step": 3150
|
2239 |
+
},
|
2240 |
+
{
|
2241 |
+
"epoch": 0.16,
|
2242 |
+
"grad_norm": 5.5,
|
2243 |
+
"learning_rate": 4.497777777777778e-07,
|
2244 |
+
"loss": 0.0581,
|
2245 |
+
"step": 3160
|
2246 |
+
},
|
2247 |
+
{
|
2248 |
+
"epoch": 0.16,
|
2249 |
+
"grad_norm": 7.21875,
|
2250 |
+
"learning_rate": 4.4733333333333334e-07,
|
2251 |
+
"loss": 0.0559,
|
2252 |
+
"step": 3170
|
2253 |
+
},
|
2254 |
+
{
|
2255 |
+
"epoch": 0.16,
|
2256 |
+
"grad_norm": 7.375,
|
2257 |
+
"learning_rate": 4.448888888888889e-07,
|
2258 |
+
"loss": 0.0537,
|
2259 |
+
"step": 3180
|
2260 |
+
},
|
2261 |
+
{
|
2262 |
+
"epoch": 0.16,
|
2263 |
+
"grad_norm": 4.75,
|
2264 |
+
"learning_rate": 4.4244444444444444e-07,
|
2265 |
+
"loss": 0.0473,
|
2266 |
+
"step": 3190
|
2267 |
+
},
|
2268 |
+
{
|
2269 |
+
"epoch": 0.16,
|
2270 |
+
"grad_norm": 7.9375,
|
2271 |
+
"learning_rate": 4.4e-07,
|
2272 |
+
"loss": 0.0494,
|
2273 |
+
"step": 3200
|
2274 |
+
},
|
2275 |
+
{
|
2276 |
+
"epoch": 0.16,
|
2277 |
+
"grad_norm": 8.5,
|
2278 |
+
"learning_rate": 4.375555555555556e-07,
|
2279 |
+
"loss": 0.058,
|
2280 |
+
"step": 3210
|
2281 |
+
},
|
2282 |
+
{
|
2283 |
+
"epoch": 0.16,
|
2284 |
+
"grad_norm": 6.0,
|
2285 |
+
"learning_rate": 4.351111111111111e-07,
|
2286 |
+
"loss": 0.057,
|
2287 |
+
"step": 3220
|
2288 |
+
},
|
2289 |
+
{
|
2290 |
+
"epoch": 0.16,
|
2291 |
+
"grad_norm": 5.125,
|
2292 |
+
"learning_rate": 4.3266666666666665e-07,
|
2293 |
+
"loss": 0.0572,
|
2294 |
+
"step": 3230
|
2295 |
+
},
|
2296 |
+
{
|
2297 |
+
"epoch": 0.16,
|
2298 |
+
"grad_norm": 5.09375,
|
2299 |
+
"learning_rate": 4.302222222222223e-07,
|
2300 |
+
"loss": 0.0525,
|
2301 |
+
"step": 3240
|
2302 |
+
},
|
2303 |
+
{
|
2304 |
+
"epoch": 0.16,
|
2305 |
+
"grad_norm": 6.8125,
|
2306 |
+
"learning_rate": 4.277777777777778e-07,
|
2307 |
+
"loss": 0.0564,
|
2308 |
+
"step": 3250
|
2309 |
+
},
|
2310 |
+
{
|
2311 |
+
"epoch": 0.16,
|
2312 |
+
"grad_norm": 6.5625,
|
2313 |
+
"learning_rate": 4.2533333333333333e-07,
|
2314 |
+
"loss": 0.0533,
|
2315 |
+
"step": 3260
|
2316 |
+
},
|
2317 |
+
{
|
2318 |
+
"epoch": 0.16,
|
2319 |
+
"grad_norm": 5.96875,
|
2320 |
+
"learning_rate": 4.228888888888889e-07,
|
2321 |
+
"loss": 0.0569,
|
2322 |
+
"step": 3270
|
2323 |
+
},
|
2324 |
+
{
|
2325 |
+
"epoch": 0.16,
|
2326 |
+
"grad_norm": 5.3125,
|
2327 |
+
"learning_rate": 4.204444444444445e-07,
|
2328 |
+
"loss": 0.0558,
|
2329 |
+
"step": 3280
|
2330 |
+
},
|
2331 |
+
{
|
2332 |
+
"epoch": 0.16,
|
2333 |
+
"grad_norm": 5.875,
|
2334 |
+
"learning_rate": 4.18e-07,
|
2335 |
+
"loss": 0.0558,
|
2336 |
+
"step": 3290
|
2337 |
+
},
|
2338 |
+
{
|
2339 |
+
"epoch": 0.17,
|
2340 |
+
"grad_norm": 6.375,
|
2341 |
+
"learning_rate": 4.155555555555556e-07,
|
2342 |
+
"loss": 0.0506,
|
2343 |
+
"step": 3300
|
2344 |
+
},
|
2345 |
+
{
|
2346 |
+
"epoch": 0.17,
|
2347 |
+
"grad_norm": 6.28125,
|
2348 |
+
"learning_rate": 4.131111111111111e-07,
|
2349 |
+
"loss": 0.0519,
|
2350 |
+
"step": 3310
|
2351 |
+
},
|
2352 |
+
{
|
2353 |
+
"epoch": 0.17,
|
2354 |
+
"grad_norm": 5.3125,
|
2355 |
+
"learning_rate": 4.106666666666667e-07,
|
2356 |
+
"loss": 0.0542,
|
2357 |
+
"step": 3320
|
2358 |
+
},
|
2359 |
+
{
|
2360 |
+
"epoch": 0.17,
|
2361 |
+
"grad_norm": 5.65625,
|
2362 |
+
"learning_rate": 4.0822222222222227e-07,
|
2363 |
+
"loss": 0.0508,
|
2364 |
+
"step": 3330
|
2365 |
+
},
|
2366 |
+
{
|
2367 |
+
"epoch": 0.17,
|
2368 |
+
"grad_norm": 8.625,
|
2369 |
+
"learning_rate": 4.057777777777778e-07,
|
2370 |
+
"loss": 0.0619,
|
2371 |
+
"step": 3340
|
2372 |
+
},
|
2373 |
+
{
|
2374 |
+
"epoch": 0.17,
|
2375 |
+
"grad_norm": 5.5,
|
2376 |
+
"learning_rate": 4.033333333333333e-07,
|
2377 |
+
"loss": 0.051,
|
2378 |
+
"step": 3350
|
2379 |
+
},
|
2380 |
+
{
|
2381 |
+
"epoch": 0.17,
|
2382 |
+
"grad_norm": 5.15625,
|
2383 |
+
"learning_rate": 4.0088888888888895e-07,
|
2384 |
+
"loss": 0.05,
|
2385 |
+
"step": 3360
|
2386 |
+
},
|
2387 |
+
{
|
2388 |
+
"epoch": 0.17,
|
2389 |
+
"grad_norm": 5.71875,
|
2390 |
+
"learning_rate": 3.984444444444445e-07,
|
2391 |
+
"loss": 0.052,
|
2392 |
+
"step": 3370
|
2393 |
+
},
|
2394 |
+
{
|
2395 |
+
"epoch": 0.17,
|
2396 |
+
"grad_norm": 7.25,
|
2397 |
+
"learning_rate": 3.96e-07,
|
2398 |
+
"loss": 0.053,
|
2399 |
+
"step": 3380
|
2400 |
+
},
|
2401 |
+
{
|
2402 |
+
"epoch": 0.17,
|
2403 |
+
"grad_norm": 6.125,
|
2404 |
+
"learning_rate": 3.935555555555555e-07,
|
2405 |
+
"loss": 0.055,
|
2406 |
+
"step": 3390
|
2407 |
+
},
|
2408 |
+
{
|
2409 |
+
"epoch": 0.17,
|
2410 |
+
"grad_norm": 5.59375,
|
2411 |
+
"learning_rate": 3.9111111111111115e-07,
|
2412 |
+
"loss": 0.0564,
|
2413 |
+
"step": 3400
|
2414 |
+
},
|
2415 |
+
{
|
2416 |
+
"epoch": 0.17,
|
2417 |
+
"grad_norm": 5.8125,
|
2418 |
+
"learning_rate": 3.886666666666667e-07,
|
2419 |
+
"loss": 0.0505,
|
2420 |
+
"step": 3410
|
2421 |
+
},
|
2422 |
+
{
|
2423 |
+
"epoch": 0.17,
|
2424 |
+
"grad_norm": 5.34375,
|
2425 |
+
"learning_rate": 3.862222222222222e-07,
|
2426 |
+
"loss": 0.048,
|
2427 |
+
"step": 3420
|
2428 |
+
},
|
2429 |
+
{
|
2430 |
+
"epoch": 0.17,
|
2431 |
+
"grad_norm": 4.9375,
|
2432 |
+
"learning_rate": 3.8377777777777783e-07,
|
2433 |
+
"loss": 0.0506,
|
2434 |
+
"step": 3430
|
2435 |
+
},
|
2436 |
+
{
|
2437 |
+
"epoch": 0.17,
|
2438 |
+
"grad_norm": 5.1875,
|
2439 |
+
"learning_rate": 3.8133333333333336e-07,
|
2440 |
+
"loss": 0.0634,
|
2441 |
+
"step": 3440
|
2442 |
+
},
|
2443 |
+
{
|
2444 |
+
"epoch": 0.17,
|
2445 |
+
"grad_norm": 8.4375,
|
2446 |
+
"learning_rate": 3.788888888888889e-07,
|
2447 |
+
"loss": 0.0546,
|
2448 |
+
"step": 3450
|
2449 |
+
},
|
2450 |
+
{
|
2451 |
+
"epoch": 0.17,
|
2452 |
+
"grad_norm": 5.625,
|
2453 |
+
"learning_rate": 3.7644444444444446e-07,
|
2454 |
+
"loss": 0.0552,
|
2455 |
+
"step": 3460
|
2456 |
+
},
|
2457 |
+
{
|
2458 |
+
"epoch": 0.17,
|
2459 |
+
"grad_norm": 6.5,
|
2460 |
+
"learning_rate": 3.7400000000000004e-07,
|
2461 |
+
"loss": 0.0512,
|
2462 |
+
"step": 3470
|
2463 |
+
},
|
2464 |
+
{
|
2465 |
+
"epoch": 0.17,
|
2466 |
+
"grad_norm": 5.8125,
|
2467 |
+
"learning_rate": 3.7155555555555557e-07,
|
2468 |
+
"loss": 0.055,
|
2469 |
+
"step": 3480
|
2470 |
+
},
|
2471 |
+
{
|
2472 |
+
"epoch": 0.17,
|
2473 |
+
"grad_norm": 5.21875,
|
2474 |
+
"learning_rate": 3.6911111111111114e-07,
|
2475 |
+
"loss": 0.0464,
|
2476 |
+
"step": 3490
|
2477 |
+
},
|
2478 |
+
{
|
2479 |
+
"epoch": 0.17,
|
2480 |
+
"grad_norm": 7.78125,
|
2481 |
+
"learning_rate": 3.6666666666666667e-07,
|
2482 |
+
"loss": 0.056,
|
2483 |
+
"step": 3500
|
2484 |
+
},
|
2485 |
+
{
|
2486 |
+
"epoch": 0.18,
|
2487 |
+
"grad_norm": 11.8125,
|
2488 |
+
"learning_rate": 3.6422222222222225e-07,
|
2489 |
+
"loss": 0.0629,
|
2490 |
+
"step": 3510
|
2491 |
+
},
|
2492 |
+
{
|
2493 |
+
"epoch": 0.18,
|
2494 |
+
"grad_norm": 5.71875,
|
2495 |
+
"learning_rate": 3.617777777777778e-07,
|
2496 |
+
"loss": 0.0595,
|
2497 |
+
"step": 3520
|
2498 |
+
},
|
2499 |
+
{
|
2500 |
+
"epoch": 0.18,
|
2501 |
+
"grad_norm": 5.75,
|
2502 |
+
"learning_rate": 3.5933333333333335e-07,
|
2503 |
+
"loss": 0.0617,
|
2504 |
+
"step": 3530
|
2505 |
+
},
|
2506 |
+
{
|
2507 |
+
"epoch": 0.18,
|
2508 |
+
"grad_norm": 5.5,
|
2509 |
+
"learning_rate": 3.5688888888888887e-07,
|
2510 |
+
"loss": 0.0556,
|
2511 |
+
"step": 3540
|
2512 |
+
},
|
2513 |
+
{
|
2514 |
+
"epoch": 0.18,
|
2515 |
+
"grad_norm": 5.90625,
|
2516 |
+
"learning_rate": 3.544444444444445e-07,
|
2517 |
+
"loss": 0.0567,
|
2518 |
+
"step": 3550
|
2519 |
+
},
|
2520 |
+
{
|
2521 |
+
"epoch": 0.18,
|
2522 |
+
"grad_norm": 6.03125,
|
2523 |
+
"learning_rate": 3.5200000000000003e-07,
|
2524 |
+
"loss": 0.0518,
|
2525 |
+
"step": 3560
|
2526 |
+
},
|
2527 |
+
{
|
2528 |
+
"epoch": 0.18,
|
2529 |
+
"grad_norm": 7.125,
|
2530 |
+
"learning_rate": 3.4955555555555555e-07,
|
2531 |
+
"loss": 0.0588,
|
2532 |
+
"step": 3570
|
2533 |
+
},
|
2534 |
+
{
|
2535 |
+
"epoch": 0.18,
|
2536 |
+
"grad_norm": 6.21875,
|
2537 |
+
"learning_rate": 3.471111111111111e-07,
|
2538 |
+
"loss": 0.0584,
|
2539 |
+
"step": 3580
|
2540 |
+
},
|
2541 |
+
{
|
2542 |
+
"epoch": 0.18,
|
2543 |
+
"grad_norm": 5.46875,
|
2544 |
+
"learning_rate": 3.446666666666667e-07,
|
2545 |
+
"loss": 0.0668,
|
2546 |
+
"step": 3590
|
2547 |
+
},
|
2548 |
+
{
|
2549 |
+
"epoch": 0.18,
|
2550 |
+
"grad_norm": 5.65625,
|
2551 |
+
"learning_rate": 3.4222222222222223e-07,
|
2552 |
+
"loss": 0.0615,
|
2553 |
+
"step": 3600
|
2554 |
+
},
|
2555 |
+
{
|
2556 |
+
"epoch": 0.18,
|
2557 |
+
"grad_norm": 5.0625,
|
2558 |
+
"learning_rate": 3.3977777777777776e-07,
|
2559 |
+
"loss": 0.0564,
|
2560 |
+
"step": 3610
|
2561 |
+
},
|
2562 |
+
{
|
2563 |
+
"epoch": 0.18,
|
2564 |
+
"grad_norm": 6.21875,
|
2565 |
+
"learning_rate": 3.3733333333333334e-07,
|
2566 |
+
"loss": 0.057,
|
2567 |
+
"step": 3620
|
2568 |
+
},
|
2569 |
+
{
|
2570 |
+
"epoch": 0.18,
|
2571 |
+
"grad_norm": 6.3125,
|
2572 |
+
"learning_rate": 3.348888888888889e-07,
|
2573 |
+
"loss": 0.0579,
|
2574 |
+
"step": 3630
|
2575 |
+
},
|
2576 |
+
{
|
2577 |
+
"epoch": 0.18,
|
2578 |
+
"grad_norm": 6.59375,
|
2579 |
+
"learning_rate": 3.3244444444444444e-07,
|
2580 |
+
"loss": 0.0577,
|
2581 |
+
"step": 3640
|
2582 |
+
},
|
2583 |
+
{
|
2584 |
+
"epoch": 0.18,
|
2585 |
+
"grad_norm": 6.59375,
|
2586 |
+
"learning_rate": 3.3e-07,
|
2587 |
+
"loss": 0.0542,
|
2588 |
+
"step": 3650
|
2589 |
+
},
|
2590 |
+
{
|
2591 |
+
"epoch": 0.18,
|
2592 |
+
"grad_norm": 5.46875,
|
2593 |
+
"learning_rate": 3.2755555555555554e-07,
|
2594 |
+
"loss": 0.0517,
|
2595 |
+
"step": 3660
|
2596 |
+
},
|
2597 |
+
{
|
2598 |
+
"epoch": 0.18,
|
2599 |
+
"grad_norm": 8.6875,
|
2600 |
+
"learning_rate": 3.251111111111111e-07,
|
2601 |
+
"loss": 0.0562,
|
2602 |
+
"step": 3670
|
2603 |
+
},
|
2604 |
+
{
|
2605 |
+
"epoch": 0.18,
|
2606 |
+
"grad_norm": 5.5625,
|
2607 |
+
"learning_rate": 3.226666666666667e-07,
|
2608 |
+
"loss": 0.0543,
|
2609 |
+
"step": 3680
|
2610 |
+
},
|
2611 |
+
{
|
2612 |
+
"epoch": 0.18,
|
2613 |
+
"grad_norm": 7.5625,
|
2614 |
+
"learning_rate": 3.202222222222222e-07,
|
2615 |
+
"loss": 0.0529,
|
2616 |
+
"step": 3690
|
2617 |
+
},
|
2618 |
+
{
|
2619 |
+
"epoch": 0.18,
|
2620 |
+
"grad_norm": 5.90625,
|
2621 |
+
"learning_rate": 3.1777777777777775e-07,
|
2622 |
+
"loss": 0.0544,
|
2623 |
+
"step": 3700
|
2624 |
+
},
|
2625 |
+
{
|
2626 |
+
"epoch": 0.19,
|
2627 |
+
"grad_norm": 7.46875,
|
2628 |
+
"learning_rate": 3.153333333333334e-07,
|
2629 |
+
"loss": 0.0519,
|
2630 |
+
"step": 3710
|
2631 |
+
},
|
2632 |
+
{
|
2633 |
+
"epoch": 0.19,
|
2634 |
+
"grad_norm": 5.5,
|
2635 |
+
"learning_rate": 3.128888888888889e-07,
|
2636 |
+
"loss": 0.0562,
|
2637 |
+
"step": 3720
|
2638 |
+
},
|
2639 |
+
{
|
2640 |
+
"epoch": 0.19,
|
2641 |
+
"grad_norm": 5.375,
|
2642 |
+
"learning_rate": 3.1044444444444443e-07,
|
2643 |
+
"loss": 0.0532,
|
2644 |
+
"step": 3730
|
2645 |
+
},
|
2646 |
+
{
|
2647 |
+
"epoch": 0.19,
|
2648 |
+
"grad_norm": 10.0,
|
2649 |
+
"learning_rate": 3.0800000000000006e-07,
|
2650 |
+
"loss": 0.0572,
|
2651 |
+
"step": 3740
|
2652 |
+
},
|
2653 |
+
{
|
2654 |
+
"epoch": 0.19,
|
2655 |
+
"grad_norm": 5.15625,
|
2656 |
+
"learning_rate": 3.055555555555556e-07,
|
2657 |
+
"loss": 0.0585,
|
2658 |
+
"step": 3750
|
2659 |
+
},
|
2660 |
+
{
|
2661 |
+
"epoch": 0.19,
|
2662 |
+
"grad_norm": 7.21875,
|
2663 |
+
"learning_rate": 3.031111111111111e-07,
|
2664 |
+
"loss": 0.0659,
|
2665 |
+
"step": 3760
|
2666 |
+
},
|
2667 |
+
{
|
2668 |
+
"epoch": 0.19,
|
2669 |
+
"grad_norm": 6.8125,
|
2670 |
+
"learning_rate": 3.006666666666667e-07,
|
2671 |
+
"loss": 0.0634,
|
2672 |
+
"step": 3770
|
2673 |
+
},
|
2674 |
+
{
|
2675 |
+
"epoch": 0.19,
|
2676 |
+
"grad_norm": 5.6875,
|
2677 |
+
"learning_rate": 2.9822222222222226e-07,
|
2678 |
+
"loss": 0.058,
|
2679 |
+
"step": 3780
|
2680 |
+
},
|
2681 |
+
{
|
2682 |
+
"epoch": 0.19,
|
2683 |
+
"grad_norm": 5.46875,
|
2684 |
+
"learning_rate": 2.957777777777778e-07,
|
2685 |
+
"loss": 0.0493,
|
2686 |
+
"step": 3790
|
2687 |
+
},
|
2688 |
+
{
|
2689 |
+
"epoch": 0.19,
|
2690 |
+
"grad_norm": 4.875,
|
2691 |
+
"learning_rate": 2.9333333333333337e-07,
|
2692 |
+
"loss": 0.0499,
|
2693 |
+
"step": 3800
|
2694 |
+
},
|
2695 |
+
{
|
2696 |
+
"epoch": 0.19,
|
2697 |
+
"grad_norm": 5.40625,
|
2698 |
+
"learning_rate": 2.908888888888889e-07,
|
2699 |
+
"loss": 0.0595,
|
2700 |
+
"step": 3810
|
2701 |
+
},
|
2702 |
+
{
|
2703 |
+
"epoch": 0.19,
|
2704 |
+
"grad_norm": 6.59375,
|
2705 |
+
"learning_rate": 2.8844444444444447e-07,
|
2706 |
+
"loss": 0.064,
|
2707 |
+
"step": 3820
|
2708 |
+
},
|
2709 |
+
{
|
2710 |
+
"epoch": 0.19,
|
2711 |
+
"grad_norm": 5.125,
|
2712 |
+
"learning_rate": 2.8600000000000005e-07,
|
2713 |
+
"loss": 0.0475,
|
2714 |
+
"step": 3830
|
2715 |
+
},
|
2716 |
+
{
|
2717 |
+
"epoch": 0.19,
|
2718 |
+
"grad_norm": 9.4375,
|
2719 |
+
"learning_rate": 2.8355555555555557e-07,
|
2720 |
+
"loss": 0.0589,
|
2721 |
+
"step": 3840
|
2722 |
+
},
|
2723 |
+
{
|
2724 |
+
"epoch": 0.19,
|
2725 |
+
"grad_norm": 5.875,
|
2726 |
+
"learning_rate": 2.811111111111111e-07,
|
2727 |
+
"loss": 0.0563,
|
2728 |
+
"step": 3850
|
2729 |
+
},
|
2730 |
+
{
|
2731 |
+
"epoch": 0.19,
|
2732 |
+
"grad_norm": 4.9375,
|
2733 |
+
"learning_rate": 2.7866666666666673e-07,
|
2734 |
+
"loss": 0.0603,
|
2735 |
+
"step": 3860
|
2736 |
+
},
|
2737 |
+
{
|
2738 |
+
"epoch": 0.19,
|
2739 |
+
"grad_norm": 5.78125,
|
2740 |
+
"learning_rate": 2.7622222222222225e-07,
|
2741 |
+
"loss": 0.0541,
|
2742 |
+
"step": 3870
|
2743 |
+
},
|
2744 |
+
{
|
2745 |
+
"epoch": 0.19,
|
2746 |
+
"grad_norm": 6.46875,
|
2747 |
+
"learning_rate": 2.737777777777778e-07,
|
2748 |
+
"loss": 0.0573,
|
2749 |
+
"step": 3880
|
2750 |
+
},
|
2751 |
+
{
|
2752 |
+
"epoch": 0.19,
|
2753 |
+
"grad_norm": 9.25,
|
2754 |
+
"learning_rate": 2.7133333333333335e-07,
|
2755 |
+
"loss": 0.0563,
|
2756 |
+
"step": 3890
|
2757 |
+
},
|
2758 |
+
{
|
2759 |
+
"epoch": 0.2,
|
2760 |
+
"grad_norm": 6.375,
|
2761 |
+
"learning_rate": 2.688888888888889e-07,
|
2762 |
+
"loss": 0.0533,
|
2763 |
+
"step": 3900
|
2764 |
+
},
|
2765 |
+
{
|
2766 |
+
"epoch": 0.2,
|
2767 |
+
"grad_norm": 9.125,
|
2768 |
+
"learning_rate": 2.6644444444444446e-07,
|
2769 |
+
"loss": 0.0528,
|
2770 |
+
"step": 3910
|
2771 |
+
},
|
2772 |
+
{
|
2773 |
+
"epoch": 0.2,
|
2774 |
+
"grad_norm": 6.71875,
|
2775 |
+
"learning_rate": 2.64e-07,
|
2776 |
+
"loss": 0.0605,
|
2777 |
+
"step": 3920
|
2778 |
+
},
|
2779 |
+
{
|
2780 |
+
"epoch": 0.2,
|
2781 |
+
"grad_norm": 6.5,
|
2782 |
+
"learning_rate": 2.6155555555555556e-07,
|
2783 |
+
"loss": 0.0605,
|
2784 |
+
"step": 3930
|
2785 |
+
},
|
2786 |
+
{
|
2787 |
+
"epoch": 0.2,
|
2788 |
+
"grad_norm": 5.34375,
|
2789 |
+
"learning_rate": 2.5911111111111114e-07,
|
2790 |
+
"loss": 0.0563,
|
2791 |
+
"step": 3940
|
2792 |
+
},
|
2793 |
+
{
|
2794 |
+
"epoch": 0.2,
|
2795 |
+
"grad_norm": 5.5625,
|
2796 |
+
"learning_rate": 2.5666666666666666e-07,
|
2797 |
+
"loss": 0.0497,
|
2798 |
+
"step": 3950
|
2799 |
+
},
|
2800 |
+
{
|
2801 |
+
"epoch": 0.2,
|
2802 |
+
"grad_norm": 9.9375,
|
2803 |
+
"learning_rate": 2.5422222222222224e-07,
|
2804 |
+
"loss": 0.0599,
|
2805 |
+
"step": 3960
|
2806 |
+
},
|
2807 |
+
{
|
2808 |
+
"epoch": 0.2,
|
2809 |
+
"grad_norm": 7.625,
|
2810 |
+
"learning_rate": 2.517777777777778e-07,
|
2811 |
+
"loss": 0.0574,
|
2812 |
+
"step": 3970
|
2813 |
+
},
|
2814 |
+
{
|
2815 |
+
"epoch": 0.2,
|
2816 |
+
"grad_norm": 6.53125,
|
2817 |
+
"learning_rate": 2.4933333333333334e-07,
|
2818 |
+
"loss": 0.0586,
|
2819 |
+
"step": 3980
|
2820 |
+
},
|
2821 |
+
{
|
2822 |
+
"epoch": 0.2,
|
2823 |
+
"grad_norm": 5.3125,
|
2824 |
+
"learning_rate": 2.468888888888889e-07,
|
2825 |
+
"loss": 0.0576,
|
2826 |
+
"step": 3990
|
2827 |
+
},
|
2828 |
+
{
|
2829 |
+
"epoch": 0.2,
|
2830 |
+
"grad_norm": 5.21875,
|
2831 |
+
"learning_rate": 2.4444444444444445e-07,
|
2832 |
+
"loss": 0.0594,
|
2833 |
+
"step": 4000
|
2834 |
+
},
|
2835 |
+
{
|
2836 |
+
"epoch": 0.2,
|
2837 |
+
"eval_loss": 0.05571923404932022,
|
2838 |
+
"eval_runtime": 64.3883,
|
2839 |
+
"eval_samples_per_second": 15.531,
|
2840 |
+
"eval_steps_per_second": 15.531,
|
2841 |
+
"step": 4000
|
2842 |
}
|
2843 |
],
|
2844 |
"logging_steps": 10,
|
|
|
2846 |
"num_input_tokens_seen": 0,
|
2847 |
"num_train_epochs": 1,
|
2848 |
"save_steps": 1000,
|
2849 |
+
"total_flos": 6.455688167424e+16,
|
2850 |
"train_batch_size": 1,
|
2851 |
"trial_name": null,
|
2852 |
"trial_params": null
|