Training checkpoint at step 7000
Browse files- trainer_state.json +366 -6
trainer_state.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
-
"best_global_step":
|
| 3 |
-
"best_metric": 2.
|
| 4 |
-
"best_model_checkpoint": "../runs/karpathy/fineweb-edu-100b-shuffle/meta-llama/Llama-3.2-1B/linear_adamw_wd1e-03_8x1024_mem32_bs64_hf_armt_dmem64/run_21/checkpoint-
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 100,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -2168,6 +2168,366 @@
|
|
| 2168 |
"eval_samples_per_second": 3.19,
|
| 2169 |
"eval_steps_per_second": 1.595,
|
| 2170 |
"step": 6000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2171 |
}
|
| 2172 |
],
|
| 2173 |
"logging_steps": 25,
|
|
@@ -2187,7 +2547,7 @@
|
|
| 2187 |
"attributes": {}
|
| 2188 |
}
|
| 2189 |
},
|
| 2190 |
-
"total_flos":
|
| 2191 |
"train_batch_size": 1,
|
| 2192 |
"trial_name": null,
|
| 2193 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_global_step": 7000,
|
| 3 |
+
"best_metric": 2.415269374847412,
|
| 4 |
+
"best_model_checkpoint": "../runs/karpathy/fineweb-edu-100b-shuffle/meta-llama/Llama-3.2-1B/linear_adamw_wd1e-03_8x1024_mem32_bs64_hf_armt_dmem64/run_21/checkpoint-7000",
|
| 5 |
+
"epoch": 0.14,
|
| 6 |
"eval_steps": 100,
|
| 7 |
+
"global_step": 7000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 2168 |
"eval_samples_per_second": 3.19,
|
| 2169 |
"eval_steps_per_second": 1.595,
|
| 2170 |
"step": 6000
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 0.1205,
|
| 2174 |
+
"grad_norm": 0.5979904516506753,
|
| 2175 |
+
"learning_rate": 9.772444444444445e-06,
|
| 2176 |
+
"loss": 2.4044,
|
| 2177 |
+
"step": 6025
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 0.121,
|
| 2181 |
+
"grad_norm": 0.5980588594331456,
|
| 2182 |
+
"learning_rate": 9.76688888888889e-06,
|
| 2183 |
+
"loss": 2.41,
|
| 2184 |
+
"step": 6050
|
| 2185 |
+
},
|
| 2186 |
+
{
|
| 2187 |
+
"epoch": 0.1215,
|
| 2188 |
+
"grad_norm": 0.6344150039672136,
|
| 2189 |
+
"learning_rate": 9.761333333333334e-06,
|
| 2190 |
+
"loss": 2.4,
|
| 2191 |
+
"step": 6075
|
| 2192 |
+
},
|
| 2193 |
+
{
|
| 2194 |
+
"epoch": 0.122,
|
| 2195 |
+
"grad_norm": 0.6035110768502723,
|
| 2196 |
+
"learning_rate": 9.755777777777778e-06,
|
| 2197 |
+
"loss": 2.4148,
|
| 2198 |
+
"step": 6100
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 0.122,
|
| 2202 |
+
"eval_loss": 2.418259382247925,
|
| 2203 |
+
"eval_runtime": 31.784,
|
| 2204 |
+
"eval_samples_per_second": 3.209,
|
| 2205 |
+
"eval_steps_per_second": 1.605,
|
| 2206 |
+
"step": 6100
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 0.1225,
|
| 2210 |
+
"grad_norm": 0.5792932239951794,
|
| 2211 |
+
"learning_rate": 9.750222222222223e-06,
|
| 2212 |
+
"loss": 2.4061,
|
| 2213 |
+
"step": 6125
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 0.123,
|
| 2217 |
+
"grad_norm": 0.6529554995007899,
|
| 2218 |
+
"learning_rate": 9.744666666666668e-06,
|
| 2219 |
+
"loss": 2.4036,
|
| 2220 |
+
"step": 6150
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 0.1235,
|
| 2224 |
+
"grad_norm": 0.5946064726146467,
|
| 2225 |
+
"learning_rate": 9.739111111111112e-06,
|
| 2226 |
+
"loss": 2.4014,
|
| 2227 |
+
"step": 6175
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 0.124,
|
| 2231 |
+
"grad_norm": 0.5739473618849045,
|
| 2232 |
+
"learning_rate": 9.733555555555555e-06,
|
| 2233 |
+
"loss": 2.4057,
|
| 2234 |
+
"step": 6200
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 0.124,
|
| 2238 |
+
"eval_loss": 2.4179208278656006,
|
| 2239 |
+
"eval_runtime": 31.6981,
|
| 2240 |
+
"eval_samples_per_second": 3.218,
|
| 2241 |
+
"eval_steps_per_second": 1.609,
|
| 2242 |
+
"step": 6200
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 0.1245,
|
| 2246 |
+
"grad_norm": 0.6907211114020956,
|
| 2247 |
+
"learning_rate": 9.728e-06,
|
| 2248 |
+
"loss": 2.393,
|
| 2249 |
+
"step": 6225
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 0.125,
|
| 2253 |
+
"grad_norm": 0.6225931887903327,
|
| 2254 |
+
"learning_rate": 9.722444444444446e-06,
|
| 2255 |
+
"loss": 2.4147,
|
| 2256 |
+
"step": 6250
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 0.1255,
|
| 2260 |
+
"grad_norm": 0.568397246680531,
|
| 2261 |
+
"learning_rate": 9.71688888888889e-06,
|
| 2262 |
+
"loss": 2.4024,
|
| 2263 |
+
"step": 6275
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 0.126,
|
| 2267 |
+
"grad_norm": 0.5842879344272728,
|
| 2268 |
+
"learning_rate": 9.711333333333333e-06,
|
| 2269 |
+
"loss": 2.404,
|
| 2270 |
+
"step": 6300
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 0.126,
|
| 2274 |
+
"eval_loss": 2.4178576469421387,
|
| 2275 |
+
"eval_runtime": 31.7994,
|
| 2276 |
+
"eval_samples_per_second": 3.208,
|
| 2277 |
+
"eval_steps_per_second": 1.604,
|
| 2278 |
+
"step": 6300
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 0.1265,
|
| 2282 |
+
"grad_norm": 0.5805192382099048,
|
| 2283 |
+
"learning_rate": 9.705777777777778e-06,
|
| 2284 |
+
"loss": 2.4063,
|
| 2285 |
+
"step": 6325
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 0.127,
|
| 2289 |
+
"grad_norm": 0.6600294122711824,
|
| 2290 |
+
"learning_rate": 9.700222222222224e-06,
|
| 2291 |
+
"loss": 2.4078,
|
| 2292 |
+
"step": 6350
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 0.1275,
|
| 2296 |
+
"grad_norm": 0.6263098682936462,
|
| 2297 |
+
"learning_rate": 9.694666666666667e-06,
|
| 2298 |
+
"loss": 2.3961,
|
| 2299 |
+
"step": 6375
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 0.128,
|
| 2303 |
+
"grad_norm": 0.6961912679129473,
|
| 2304 |
+
"learning_rate": 9.68911111111111e-06,
|
| 2305 |
+
"loss": 2.4127,
|
| 2306 |
+
"step": 6400
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 0.128,
|
| 2310 |
+
"eval_loss": 2.417247772216797,
|
| 2311 |
+
"eval_runtime": 31.7325,
|
| 2312 |
+
"eval_samples_per_second": 3.214,
|
| 2313 |
+
"eval_steps_per_second": 1.607,
|
| 2314 |
+
"step": 6400
|
| 2315 |
+
},
|
| 2316 |
+
{
|
| 2317 |
+
"epoch": 0.1285,
|
| 2318 |
+
"grad_norm": 0.6396950069271417,
|
| 2319 |
+
"learning_rate": 9.683555555555556e-06,
|
| 2320 |
+
"loss": 2.4041,
|
| 2321 |
+
"step": 6425
|
| 2322 |
+
},
|
| 2323 |
+
{
|
| 2324 |
+
"epoch": 0.129,
|
| 2325 |
+
"grad_norm": 0.6164180606933177,
|
| 2326 |
+
"learning_rate": 9.678000000000001e-06,
|
| 2327 |
+
"loss": 2.4,
|
| 2328 |
+
"step": 6450
|
| 2329 |
+
},
|
| 2330 |
+
{
|
| 2331 |
+
"epoch": 0.1295,
|
| 2332 |
+
"grad_norm": 0.6120640198257105,
|
| 2333 |
+
"learning_rate": 9.672444444444445e-06,
|
| 2334 |
+
"loss": 2.3966,
|
| 2335 |
+
"step": 6475
|
| 2336 |
+
},
|
| 2337 |
+
{
|
| 2338 |
+
"epoch": 0.13,
|
| 2339 |
+
"grad_norm": 0.6013045247718226,
|
| 2340 |
+
"learning_rate": 9.66688888888889e-06,
|
| 2341 |
+
"loss": 2.3991,
|
| 2342 |
+
"step": 6500
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 0.13,
|
| 2346 |
+
"eval_loss": 2.417280673980713,
|
| 2347 |
+
"eval_runtime": 31.8112,
|
| 2348 |
+
"eval_samples_per_second": 3.206,
|
| 2349 |
+
"eval_steps_per_second": 1.603,
|
| 2350 |
+
"step": 6500
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 0.1305,
|
| 2354 |
+
"grad_norm": 0.6061836537875764,
|
| 2355 |
+
"learning_rate": 9.661333333333334e-06,
|
| 2356 |
+
"loss": 2.4161,
|
| 2357 |
+
"step": 6525
|
| 2358 |
+
},
|
| 2359 |
+
{
|
| 2360 |
+
"epoch": 0.131,
|
| 2361 |
+
"grad_norm": 0.6100864625060891,
|
| 2362 |
+
"learning_rate": 9.655777777777779e-06,
|
| 2363 |
+
"loss": 2.4052,
|
| 2364 |
+
"step": 6550
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 0.1315,
|
| 2368 |
+
"grad_norm": 0.6932893052541476,
|
| 2369 |
+
"learning_rate": 9.650222222222222e-06,
|
| 2370 |
+
"loss": 2.4036,
|
| 2371 |
+
"step": 6575
|
| 2372 |
+
},
|
| 2373 |
+
{
|
| 2374 |
+
"epoch": 0.132,
|
| 2375 |
+
"grad_norm": 0.5859072202807338,
|
| 2376 |
+
"learning_rate": 9.644666666666668e-06,
|
| 2377 |
+
"loss": 2.4045,
|
| 2378 |
+
"step": 6600
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 0.132,
|
| 2382 |
+
"eval_loss": 2.416877031326294,
|
| 2383 |
+
"eval_runtime": 31.5203,
|
| 2384 |
+
"eval_samples_per_second": 3.236,
|
| 2385 |
+
"eval_steps_per_second": 1.618,
|
| 2386 |
+
"step": 6600
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 0.1325,
|
| 2390 |
+
"grad_norm": 0.579002436095642,
|
| 2391 |
+
"learning_rate": 9.639111111111113e-06,
|
| 2392 |
+
"loss": 2.4015,
|
| 2393 |
+
"step": 6625
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 0.133,
|
| 2397 |
+
"grad_norm": 0.5968858601649685,
|
| 2398 |
+
"learning_rate": 9.633555555555556e-06,
|
| 2399 |
+
"loss": 2.3986,
|
| 2400 |
+
"step": 6650
|
| 2401 |
+
},
|
| 2402 |
+
{
|
| 2403 |
+
"epoch": 0.1335,
|
| 2404 |
+
"grad_norm": 0.5964714549861985,
|
| 2405 |
+
"learning_rate": 9.628e-06,
|
| 2406 |
+
"loss": 2.4062,
|
| 2407 |
+
"step": 6675
|
| 2408 |
+
},
|
| 2409 |
+
{
|
| 2410 |
+
"epoch": 0.134,
|
| 2411 |
+
"grad_norm": 0.6126102944808797,
|
| 2412 |
+
"learning_rate": 9.622444444444445e-06,
|
| 2413 |
+
"loss": 2.4033,
|
| 2414 |
+
"step": 6700
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 0.134,
|
| 2418 |
+
"eval_loss": 2.4164350032806396,
|
| 2419 |
+
"eval_runtime": 31.4543,
|
| 2420 |
+
"eval_samples_per_second": 3.243,
|
| 2421 |
+
"eval_steps_per_second": 1.621,
|
| 2422 |
+
"step": 6700
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 0.1345,
|
| 2426 |
+
"grad_norm": 0.5774452345333466,
|
| 2427 |
+
"learning_rate": 9.61688888888889e-06,
|
| 2428 |
+
"loss": 2.3997,
|
| 2429 |
+
"step": 6725
|
| 2430 |
+
},
|
| 2431 |
+
{
|
| 2432 |
+
"epoch": 0.135,
|
| 2433 |
+
"grad_norm": 0.6227260743975279,
|
| 2434 |
+
"learning_rate": 9.611333333333334e-06,
|
| 2435 |
+
"loss": 2.4018,
|
| 2436 |
+
"step": 6750
|
| 2437 |
+
},
|
| 2438 |
+
{
|
| 2439 |
+
"epoch": 0.1355,
|
| 2440 |
+
"grad_norm": 0.5846707991616706,
|
| 2441 |
+
"learning_rate": 9.605777777777778e-06,
|
| 2442 |
+
"loss": 2.3985,
|
| 2443 |
+
"step": 6775
|
| 2444 |
+
},
|
| 2445 |
+
{
|
| 2446 |
+
"epoch": 0.136,
|
| 2447 |
+
"grad_norm": 0.6172483484063671,
|
| 2448 |
+
"learning_rate": 9.600222222222223e-06,
|
| 2449 |
+
"loss": 2.4213,
|
| 2450 |
+
"step": 6800
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 0.136,
|
| 2454 |
+
"eval_loss": 2.41625714302063,
|
| 2455 |
+
"eval_runtime": 31.5517,
|
| 2456 |
+
"eval_samples_per_second": 3.233,
|
| 2457 |
+
"eval_steps_per_second": 1.616,
|
| 2458 |
+
"step": 6800
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 0.1365,
|
| 2462 |
+
"grad_norm": 0.5965299711032601,
|
| 2463 |
+
"learning_rate": 9.594666666666668e-06,
|
| 2464 |
+
"loss": 2.3976,
|
| 2465 |
+
"step": 6825
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 0.137,
|
| 2469 |
+
"grad_norm": 0.5884739304234496,
|
| 2470 |
+
"learning_rate": 9.589111111111112e-06,
|
| 2471 |
+
"loss": 2.3947,
|
| 2472 |
+
"step": 6850
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 0.1375,
|
| 2476 |
+
"grad_norm": 0.5737065693146471,
|
| 2477 |
+
"learning_rate": 9.583555555555555e-06,
|
| 2478 |
+
"loss": 2.3983,
|
| 2479 |
+
"step": 6875
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 0.138,
|
| 2483 |
+
"grad_norm": 0.6249698819825935,
|
| 2484 |
+
"learning_rate": 9.578e-06,
|
| 2485 |
+
"loss": 2.4008,
|
| 2486 |
+
"step": 6900
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 0.138,
|
| 2490 |
+
"eval_loss": 2.4156551361083984,
|
| 2491 |
+
"eval_runtime": 31.5071,
|
| 2492 |
+
"eval_samples_per_second": 3.237,
|
| 2493 |
+
"eval_steps_per_second": 1.619,
|
| 2494 |
+
"step": 6900
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 0.1385,
|
| 2498 |
+
"grad_norm": 0.5930008566650997,
|
| 2499 |
+
"learning_rate": 9.572444444444446e-06,
|
| 2500 |
+
"loss": 2.3951,
|
| 2501 |
+
"step": 6925
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 0.139,
|
| 2505 |
+
"grad_norm": 0.6564746022716046,
|
| 2506 |
+
"learning_rate": 9.56688888888889e-06,
|
| 2507 |
+
"loss": 2.4083,
|
| 2508 |
+
"step": 6950
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 0.1395,
|
| 2512 |
+
"grad_norm": 0.611311960098376,
|
| 2513 |
+
"learning_rate": 9.561333333333333e-06,
|
| 2514 |
+
"loss": 2.4032,
|
| 2515 |
+
"step": 6975
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 0.14,
|
| 2519 |
+
"grad_norm": 0.594692534551516,
|
| 2520 |
+
"learning_rate": 9.555777777777778e-06,
|
| 2521 |
+
"loss": 2.41,
|
| 2522 |
+
"step": 7000
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 0.14,
|
| 2526 |
+
"eval_loss": 2.415269374847412,
|
| 2527 |
+
"eval_runtime": 31.7535,
|
| 2528 |
+
"eval_samples_per_second": 3.212,
|
| 2529 |
+
"eval_steps_per_second": 1.606,
|
| 2530 |
+
"step": 7000
|
| 2531 |
}
|
| 2532 |
],
|
| 2533 |
"logging_steps": 25,
|
|
|
|
| 2547 |
"attributes": {}
|
| 2548 |
}
|
| 2549 |
},
|
| 2550 |
+
"total_flos": 2.2282416088290427e+19,
|
| 2551 |
"train_batch_size": 1,
|
| 2552 |
"trial_name": null,
|
| 2553 |
"trial_params": null
|