Training in progress, step 4000, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 598434152
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5fbedc25a7e866c5375d17390d5eb67e93e519d92626a3f97c7daddf7702fcab
|
| 3 |
size 598434152
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1196955962
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ff74a5fb5071ae21ddeb0a899cb40c982a667d261c8d2108985988527a0acde6
|
| 3 |
size 1196955962
|
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 sha256:8f9aa43992237c34da047eae3e4635545e3cbee9026436669a8ec61ef48f58c1
|
| 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 1000
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4dc07d6a74a610820a65e18607f99e43ef207c98dc79fcbe54bd9e13a24f2410
|
| 3 |
size 1000
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 1.
|
| 5 |
"eval_steps": 250,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -2299,6 +2299,770 @@
|
|
| 2299 |
"eval_spearman_manhattan": 0.7832975474858643,
|
| 2300 |
"eval_steps_per_second": 31.371,
|
| 2301 |
"step": 3000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2302 |
}
|
| 2303 |
],
|
| 2304 |
"logging_steps": 10,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.8744142455482662,
|
| 5 |
"eval_steps": 250,
|
| 6 |
+
"global_step": 4000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2299 |
"eval_spearman_manhattan": 0.7832975474858643,
|
| 2300 |
"eval_steps_per_second": 31.371,
|
| 2301 |
"step": 3000
|
| 2302 |
+
},
|
| 2303 |
+
{
|
| 2304 |
+
"epoch": 1.4104967197750704,
|
| 2305 |
+
"grad_norm": 4.2792487144470215,
|
| 2306 |
+
"learning_rate": 9.118439550140582e-06,
|
| 2307 |
+
"loss": 0.4878,
|
| 2308 |
+
"step": 3010
|
| 2309 |
+
},
|
| 2310 |
+
{
|
| 2311 |
+
"epoch": 1.415182755388941,
|
| 2312 |
+
"grad_norm": 3.8892383575439453,
|
| 2313 |
+
"learning_rate": 9.115510777881913e-06,
|
| 2314 |
+
"loss": 0.4676,
|
| 2315 |
+
"step": 3020
|
| 2316 |
+
},
|
| 2317 |
+
{
|
| 2318 |
+
"epoch": 1.4198687910028116,
|
| 2319 |
+
"grad_norm": 5.0008745193481445,
|
| 2320 |
+
"learning_rate": 9.112582005623244e-06,
|
| 2321 |
+
"loss": 0.4729,
|
| 2322 |
+
"step": 3030
|
| 2323 |
+
},
|
| 2324 |
+
{
|
| 2325 |
+
"epoch": 1.4245548266166823,
|
| 2326 |
+
"grad_norm": 5.607409477233887,
|
| 2327 |
+
"learning_rate": 9.109653233364575e-06,
|
| 2328 |
+
"loss": 0.4762,
|
| 2329 |
+
"step": 3040
|
| 2330 |
+
},
|
| 2331 |
+
{
|
| 2332 |
+
"epoch": 1.429240862230553,
|
| 2333 |
+
"grad_norm": 3.0340139865875244,
|
| 2334 |
+
"learning_rate": 9.106724461105905e-06,
|
| 2335 |
+
"loss": 0.4438,
|
| 2336 |
+
"step": 3050
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 1.4339268978444237,
|
| 2340 |
+
"grad_norm": 4.310724258422852,
|
| 2341 |
+
"learning_rate": 9.103795688847236e-06,
|
| 2342 |
+
"loss": 0.4499,
|
| 2343 |
+
"step": 3060
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 1.4386129334582942,
|
| 2347 |
+
"grad_norm": 4.481917381286621,
|
| 2348 |
+
"learning_rate": 9.100866916588567e-06,
|
| 2349 |
+
"loss": 0.4493,
|
| 2350 |
+
"step": 3070
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 1.443298969072165,
|
| 2354 |
+
"grad_norm": 4.330621719360352,
|
| 2355 |
+
"learning_rate": 9.097938144329898e-06,
|
| 2356 |
+
"loss": 0.4505,
|
| 2357 |
+
"step": 3080
|
| 2358 |
+
},
|
| 2359 |
+
{
|
| 2360 |
+
"epoch": 1.4479850046860356,
|
| 2361 |
+
"grad_norm": 4.335081577301025,
|
| 2362 |
+
"learning_rate": 9.095009372071228e-06,
|
| 2363 |
+
"loss": 0.446,
|
| 2364 |
+
"step": 3090
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 1.4526710402999063,
|
| 2368 |
+
"grad_norm": 3.0894672870635986,
|
| 2369 |
+
"learning_rate": 9.092080599812559e-06,
|
| 2370 |
+
"loss": 0.4404,
|
| 2371 |
+
"step": 3100
|
| 2372 |
+
},
|
| 2373 |
+
{
|
| 2374 |
+
"epoch": 1.457357075913777,
|
| 2375 |
+
"grad_norm": 4.6363983154296875,
|
| 2376 |
+
"learning_rate": 9.08915182755389e-06,
|
| 2377 |
+
"loss": 0.5358,
|
| 2378 |
+
"step": 3110
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 1.4620431115276475,
|
| 2382 |
+
"grad_norm": 3.80387806892395,
|
| 2383 |
+
"learning_rate": 9.086223055295221e-06,
|
| 2384 |
+
"loss": 0.4374,
|
| 2385 |
+
"step": 3120
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 1.4667291471415183,
|
| 2389 |
+
"grad_norm": 3.276442289352417,
|
| 2390 |
+
"learning_rate": 9.083294283036552e-06,
|
| 2391 |
+
"loss": 0.5013,
|
| 2392 |
+
"step": 3130
|
| 2393 |
+
},
|
| 2394 |
+
{
|
| 2395 |
+
"epoch": 1.471415182755389,
|
| 2396 |
+
"grad_norm": 3.843419075012207,
|
| 2397 |
+
"learning_rate": 9.080365510777884e-06,
|
| 2398 |
+
"loss": 0.4694,
|
| 2399 |
+
"step": 3140
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 1.4761012183692597,
|
| 2403 |
+
"grad_norm": 4.7606730461120605,
|
| 2404 |
+
"learning_rate": 9.077436738519213e-06,
|
| 2405 |
+
"loss": 0.4215,
|
| 2406 |
+
"step": 3150
|
| 2407 |
+
},
|
| 2408 |
+
{
|
| 2409 |
+
"epoch": 1.4807872539831304,
|
| 2410 |
+
"grad_norm": 3.739225149154663,
|
| 2411 |
+
"learning_rate": 9.074507966260544e-06,
|
| 2412 |
+
"loss": 0.4756,
|
| 2413 |
+
"step": 3160
|
| 2414 |
+
},
|
| 2415 |
+
{
|
| 2416 |
+
"epoch": 1.4854732895970009,
|
| 2417 |
+
"grad_norm": 3.36938214302063,
|
| 2418 |
+
"learning_rate": 9.071579194001876e-06,
|
| 2419 |
+
"loss": 0.4243,
|
| 2420 |
+
"step": 3170
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 1.4901593252108716,
|
| 2424 |
+
"grad_norm": 6.589993476867676,
|
| 2425 |
+
"learning_rate": 9.068650421743205e-06,
|
| 2426 |
+
"loss": 0.4698,
|
| 2427 |
+
"step": 3180
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 1.4948453608247423,
|
| 2431 |
+
"grad_norm": 3.8416695594787598,
|
| 2432 |
+
"learning_rate": 9.065721649484536e-06,
|
| 2433 |
+
"loss": 0.4964,
|
| 2434 |
+
"step": 3190
|
| 2435 |
+
},
|
| 2436 |
+
{
|
| 2437 |
+
"epoch": 1.499531396438613,
|
| 2438 |
+
"grad_norm": 4.367741584777832,
|
| 2439 |
+
"learning_rate": 9.062792877225867e-06,
|
| 2440 |
+
"loss": 0.4417,
|
| 2441 |
+
"step": 3200
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 1.5042174320524837,
|
| 2445 |
+
"grad_norm": 3.500617742538452,
|
| 2446 |
+
"learning_rate": 9.059864104967199e-06,
|
| 2447 |
+
"loss": 0.4522,
|
| 2448 |
+
"step": 3210
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 1.5089034676663542,
|
| 2452 |
+
"grad_norm": 3.5349769592285156,
|
| 2453 |
+
"learning_rate": 9.05693533270853e-06,
|
| 2454 |
+
"loss": 0.4393,
|
| 2455 |
+
"step": 3220
|
| 2456 |
+
},
|
| 2457 |
+
{
|
| 2458 |
+
"epoch": 1.513589503280225,
|
| 2459 |
+
"grad_norm": 3.8469526767730713,
|
| 2460 |
+
"learning_rate": 9.054006560449861e-06,
|
| 2461 |
+
"loss": 0.4453,
|
| 2462 |
+
"step": 3230
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 1.5182755388940956,
|
| 2466 |
+
"grad_norm": 3.209933280944824,
|
| 2467 |
+
"learning_rate": 9.051077788191192e-06,
|
| 2468 |
+
"loss": 0.4599,
|
| 2469 |
+
"step": 3240
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 1.522961574507966,
|
| 2473 |
+
"grad_norm": 3.7976036071777344,
|
| 2474 |
+
"learning_rate": 9.048149015932522e-06,
|
| 2475 |
+
"loss": 0.4373,
|
| 2476 |
+
"step": 3250
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 1.522961574507966,
|
| 2480 |
+
"eval_loss": 0.049798864871263504,
|
| 2481 |
+
"eval_pearson_cosine": 0.7866421286010308,
|
| 2482 |
+
"eval_pearson_dot": 0.6641640853451243,
|
| 2483 |
+
"eval_pearson_euclidean": 0.7777378719378305,
|
| 2484 |
+
"eval_pearson_manhattan": 0.7764827785285746,
|
| 2485 |
+
"eval_runtime": 43.7509,
|
| 2486 |
+
"eval_samples_per_second": 34.285,
|
| 2487 |
+
"eval_spearman_cosine": 0.7870351053050699,
|
| 2488 |
+
"eval_spearman_dot": 0.6708598238937284,
|
| 2489 |
+
"eval_spearman_euclidean": 0.7874683707378692,
|
| 2490 |
+
"eval_spearman_manhattan": 0.7865203522698128,
|
| 2491 |
+
"eval_steps_per_second": 34.285,
|
| 2492 |
+
"step": 3250
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 1.527647610121837,
|
| 2496 |
+
"grad_norm": 4.851262092590332,
|
| 2497 |
+
"learning_rate": 9.045220243673853e-06,
|
| 2498 |
+
"loss": 0.491,
|
| 2499 |
+
"step": 3260
|
| 2500 |
+
},
|
| 2501 |
+
{
|
| 2502 |
+
"epoch": 1.5323336457357075,
|
| 2503 |
+
"grad_norm": 4.183891773223877,
|
| 2504 |
+
"learning_rate": 9.042291471415184e-06,
|
| 2505 |
+
"loss": 0.453,
|
| 2506 |
+
"step": 3270
|
| 2507 |
+
},
|
| 2508 |
+
{
|
| 2509 |
+
"epoch": 1.5370196813495782,
|
| 2510 |
+
"grad_norm": 4.280774116516113,
|
| 2511 |
+
"learning_rate": 9.039362699156514e-06,
|
| 2512 |
+
"loss": 0.4413,
|
| 2513 |
+
"step": 3280
|
| 2514 |
+
},
|
| 2515 |
+
{
|
| 2516 |
+
"epoch": 1.541705716963449,
|
| 2517 |
+
"grad_norm": 4.118307113647461,
|
| 2518 |
+
"learning_rate": 9.036433926897845e-06,
|
| 2519 |
+
"loss": 0.4661,
|
| 2520 |
+
"step": 3290
|
| 2521 |
+
},
|
| 2522 |
+
{
|
| 2523 |
+
"epoch": 1.5463917525773194,
|
| 2524 |
+
"grad_norm": 5.99712610244751,
|
| 2525 |
+
"learning_rate": 9.033505154639176e-06,
|
| 2526 |
+
"loss": 0.5205,
|
| 2527 |
+
"step": 3300
|
| 2528 |
+
},
|
| 2529 |
+
{
|
| 2530 |
+
"epoch": 1.5510777881911904,
|
| 2531 |
+
"grad_norm": 4.146691799163818,
|
| 2532 |
+
"learning_rate": 9.030576382380507e-06,
|
| 2533 |
+
"loss": 0.428,
|
| 2534 |
+
"step": 3310
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 1.5557638238050608,
|
| 2538 |
+
"grad_norm": 3.899887800216675,
|
| 2539 |
+
"learning_rate": 9.027647610121838e-06,
|
| 2540 |
+
"loss": 0.4564,
|
| 2541 |
+
"step": 3320
|
| 2542 |
+
},
|
| 2543 |
+
{
|
| 2544 |
+
"epoch": 1.5604498594189316,
|
| 2545 |
+
"grad_norm": 3.9663302898406982,
|
| 2546 |
+
"learning_rate": 9.02471883786317e-06,
|
| 2547 |
+
"loss": 0.4539,
|
| 2548 |
+
"step": 3330
|
| 2549 |
+
},
|
| 2550 |
+
{
|
| 2551 |
+
"epoch": 1.5651358950328023,
|
| 2552 |
+
"grad_norm": 3.526458263397217,
|
| 2553 |
+
"learning_rate": 9.021790065604499e-06,
|
| 2554 |
+
"loss": 0.4844,
|
| 2555 |
+
"step": 3340
|
| 2556 |
+
},
|
| 2557 |
+
{
|
| 2558 |
+
"epoch": 1.569821930646673,
|
| 2559 |
+
"grad_norm": 4.192911624908447,
|
| 2560 |
+
"learning_rate": 9.01886129334583e-06,
|
| 2561 |
+
"loss": 0.4278,
|
| 2562 |
+
"step": 3350
|
| 2563 |
+
},
|
| 2564 |
+
{
|
| 2565 |
+
"epoch": 1.5745079662605437,
|
| 2566 |
+
"grad_norm": 4.185749530792236,
|
| 2567 |
+
"learning_rate": 9.015932521087161e-06,
|
| 2568 |
+
"loss": 0.4632,
|
| 2569 |
+
"step": 3360
|
| 2570 |
+
},
|
| 2571 |
+
{
|
| 2572 |
+
"epoch": 1.5791940018744142,
|
| 2573 |
+
"grad_norm": 3.411773204803467,
|
| 2574 |
+
"learning_rate": 9.013003748828491e-06,
|
| 2575 |
+
"loss": 0.436,
|
| 2576 |
+
"step": 3370
|
| 2577 |
+
},
|
| 2578 |
+
{
|
| 2579 |
+
"epoch": 1.5838800374882849,
|
| 2580 |
+
"grad_norm": 4.467881679534912,
|
| 2581 |
+
"learning_rate": 9.010074976569822e-06,
|
| 2582 |
+
"loss": 0.4133,
|
| 2583 |
+
"step": 3380
|
| 2584 |
+
},
|
| 2585 |
+
{
|
| 2586 |
+
"epoch": 1.5885660731021556,
|
| 2587 |
+
"grad_norm": 3.77736496925354,
|
| 2588 |
+
"learning_rate": 9.007146204311153e-06,
|
| 2589 |
+
"loss": 0.4452,
|
| 2590 |
+
"step": 3390
|
| 2591 |
+
},
|
| 2592 |
+
{
|
| 2593 |
+
"epoch": 1.5932521087160263,
|
| 2594 |
+
"grad_norm": 4.084095478057861,
|
| 2595 |
+
"learning_rate": 9.004217432052485e-06,
|
| 2596 |
+
"loss": 0.4605,
|
| 2597 |
+
"step": 3400
|
| 2598 |
+
},
|
| 2599 |
+
{
|
| 2600 |
+
"epoch": 1.597938144329897,
|
| 2601 |
+
"grad_norm": 3.3393008708953857,
|
| 2602 |
+
"learning_rate": 9.001288659793816e-06,
|
| 2603 |
+
"loss": 0.4157,
|
| 2604 |
+
"step": 3410
|
| 2605 |
+
},
|
| 2606 |
+
{
|
| 2607 |
+
"epoch": 1.6026241799437675,
|
| 2608 |
+
"grad_norm": 3.096881151199341,
|
| 2609 |
+
"learning_rate": 8.998359887535147e-06,
|
| 2610 |
+
"loss": 0.4478,
|
| 2611 |
+
"step": 3420
|
| 2612 |
+
},
|
| 2613 |
+
{
|
| 2614 |
+
"epoch": 1.6073102155576382,
|
| 2615 |
+
"grad_norm": 3.0557243824005127,
|
| 2616 |
+
"learning_rate": 8.995431115276478e-06,
|
| 2617 |
+
"loss": 0.4452,
|
| 2618 |
+
"step": 3430
|
| 2619 |
+
},
|
| 2620 |
+
{
|
| 2621 |
+
"epoch": 1.611996251171509,
|
| 2622 |
+
"grad_norm": 3.7997219562530518,
|
| 2623 |
+
"learning_rate": 8.992502343017808e-06,
|
| 2624 |
+
"loss": 0.4287,
|
| 2625 |
+
"step": 3440
|
| 2626 |
+
},
|
| 2627 |
+
{
|
| 2628 |
+
"epoch": 1.6166822867853796,
|
| 2629 |
+
"grad_norm": 3.6995465755462646,
|
| 2630 |
+
"learning_rate": 8.989573570759139e-06,
|
| 2631 |
+
"loss": 0.4423,
|
| 2632 |
+
"step": 3450
|
| 2633 |
+
},
|
| 2634 |
+
{
|
| 2635 |
+
"epoch": 1.6213683223992503,
|
| 2636 |
+
"grad_norm": 4.1384053230285645,
|
| 2637 |
+
"learning_rate": 8.986644798500468e-06,
|
| 2638 |
+
"loss": 0.4563,
|
| 2639 |
+
"step": 3460
|
| 2640 |
+
},
|
| 2641 |
+
{
|
| 2642 |
+
"epoch": 1.6260543580131208,
|
| 2643 |
+
"grad_norm": 4.637014865875244,
|
| 2644 |
+
"learning_rate": 8.9837160262418e-06,
|
| 2645 |
+
"loss": 0.4538,
|
| 2646 |
+
"step": 3470
|
| 2647 |
+
},
|
| 2648 |
+
{
|
| 2649 |
+
"epoch": 1.6307403936269915,
|
| 2650 |
+
"grad_norm": 4.30952262878418,
|
| 2651 |
+
"learning_rate": 8.98078725398313e-06,
|
| 2652 |
+
"loss": 0.3993,
|
| 2653 |
+
"step": 3480
|
| 2654 |
+
},
|
| 2655 |
+
{
|
| 2656 |
+
"epoch": 1.6354264292408622,
|
| 2657 |
+
"grad_norm": 4.746737003326416,
|
| 2658 |
+
"learning_rate": 8.977858481724462e-06,
|
| 2659 |
+
"loss": 0.4274,
|
| 2660 |
+
"step": 3490
|
| 2661 |
+
},
|
| 2662 |
+
{
|
| 2663 |
+
"epoch": 1.640112464854733,
|
| 2664 |
+
"grad_norm": 3.8592286109924316,
|
| 2665 |
+
"learning_rate": 8.974929709465793e-06,
|
| 2666 |
+
"loss": 0.4066,
|
| 2667 |
+
"step": 3500
|
| 2668 |
+
},
|
| 2669 |
+
{
|
| 2670 |
+
"epoch": 1.640112464854733,
|
| 2671 |
+
"eval_loss": 0.050406068563461304,
|
| 2672 |
+
"eval_pearson_cosine": 0.7840015528942317,
|
| 2673 |
+
"eval_pearson_dot": 0.659932129633507,
|
| 2674 |
+
"eval_pearson_euclidean": 0.7769297052026758,
|
| 2675 |
+
"eval_pearson_manhattan": 0.7754185185705609,
|
| 2676 |
+
"eval_runtime": 44.0859,
|
| 2677 |
+
"eval_samples_per_second": 34.024,
|
| 2678 |
+
"eval_spearman_cosine": 0.7845451302239834,
|
| 2679 |
+
"eval_spearman_dot": 0.6667296644451466,
|
| 2680 |
+
"eval_spearman_euclidean": 0.7868327314956118,
|
| 2681 |
+
"eval_spearman_manhattan": 0.7856021398727839,
|
| 2682 |
+
"eval_steps_per_second": 34.024,
|
| 2683 |
+
"step": 3500
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 1.6447985004686037,
|
| 2687 |
+
"grad_norm": 5.983098030090332,
|
| 2688 |
+
"learning_rate": 8.972000937207124e-06,
|
| 2689 |
+
"loss": 0.4451,
|
| 2690 |
+
"step": 3510
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 1.6494845360824741,
|
| 2694 |
+
"grad_norm": 4.052550315856934,
|
| 2695 |
+
"learning_rate": 8.969072164948455e-06,
|
| 2696 |
+
"loss": 0.4331,
|
| 2697 |
+
"step": 3520
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 1.6541705716963448,
|
| 2701 |
+
"grad_norm": 3.7970380783081055,
|
| 2702 |
+
"learning_rate": 8.966143392689785e-06,
|
| 2703 |
+
"loss": 0.4427,
|
| 2704 |
+
"step": 3530
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"epoch": 1.6588566073102156,
|
| 2708 |
+
"grad_norm": 4.695807456970215,
|
| 2709 |
+
"learning_rate": 8.963214620431116e-06,
|
| 2710 |
+
"loss": 0.4522,
|
| 2711 |
+
"step": 3540
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 1.6635426429240863,
|
| 2715 |
+
"grad_norm": 4.41202974319458,
|
| 2716 |
+
"learning_rate": 8.960285848172446e-06,
|
| 2717 |
+
"loss": 0.4275,
|
| 2718 |
+
"step": 3550
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 1.668228678537957,
|
| 2722 |
+
"grad_norm": 5.364877223968506,
|
| 2723 |
+
"learning_rate": 8.957357075913777e-06,
|
| 2724 |
+
"loss": 0.4321,
|
| 2725 |
+
"step": 3560
|
| 2726 |
+
},
|
| 2727 |
+
{
|
| 2728 |
+
"epoch": 1.6729147141518275,
|
| 2729 |
+
"grad_norm": 3.801132917404175,
|
| 2730 |
+
"learning_rate": 8.954428303655108e-06,
|
| 2731 |
+
"loss": 0.4494,
|
| 2732 |
+
"step": 3570
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 1.6776007497656982,
|
| 2736 |
+
"grad_norm": 4.197866439819336,
|
| 2737 |
+
"learning_rate": 8.95149953139644e-06,
|
| 2738 |
+
"loss": 0.4126,
|
| 2739 |
+
"step": 3580
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 1.6822867853795689,
|
| 2743 |
+
"grad_norm": 5.34595251083374,
|
| 2744 |
+
"learning_rate": 8.94857075913777e-06,
|
| 2745 |
+
"loss": 0.4757,
|
| 2746 |
+
"step": 3590
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 1.6869728209934396,
|
| 2750 |
+
"grad_norm": 4.772789478302002,
|
| 2751 |
+
"learning_rate": 8.945641986879102e-06,
|
| 2752 |
+
"loss": 0.4037,
|
| 2753 |
+
"step": 3600
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 1.6916588566073103,
|
| 2757 |
+
"grad_norm": 4.81839656829834,
|
| 2758 |
+
"learning_rate": 8.942713214620433e-06,
|
| 2759 |
+
"loss": 0.4192,
|
| 2760 |
+
"step": 3610
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 1.6963448922211808,
|
| 2764 |
+
"grad_norm": 3.470919132232666,
|
| 2765 |
+
"learning_rate": 8.939784442361762e-06,
|
| 2766 |
+
"loss": 0.4106,
|
| 2767 |
+
"step": 3620
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 1.7010309278350515,
|
| 2771 |
+
"grad_norm": 3.2051522731781006,
|
| 2772 |
+
"learning_rate": 8.936855670103094e-06,
|
| 2773 |
+
"loss": 0.4162,
|
| 2774 |
+
"step": 3630
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 1.7057169634489222,
|
| 2778 |
+
"grad_norm": 3.8122334480285645,
|
| 2779 |
+
"learning_rate": 8.933926897844423e-06,
|
| 2780 |
+
"loss": 0.4054,
|
| 2781 |
+
"step": 3640
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 1.710402999062793,
|
| 2785 |
+
"grad_norm": 5.07956075668335,
|
| 2786 |
+
"learning_rate": 8.930998125585754e-06,
|
| 2787 |
+
"loss": 0.4164,
|
| 2788 |
+
"step": 3650
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 1.7150890346766636,
|
| 2792 |
+
"grad_norm": 3.754542112350464,
|
| 2793 |
+
"learning_rate": 8.928069353327085e-06,
|
| 2794 |
+
"loss": 0.3703,
|
| 2795 |
+
"step": 3660
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 1.7197750702905341,
|
| 2799 |
+
"grad_norm": 3.4620890617370605,
|
| 2800 |
+
"learning_rate": 8.925140581068417e-06,
|
| 2801 |
+
"loss": 0.4667,
|
| 2802 |
+
"step": 3670
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 1.7244611059044048,
|
| 2806 |
+
"grad_norm": 4.179393768310547,
|
| 2807 |
+
"learning_rate": 8.922211808809748e-06,
|
| 2808 |
+
"loss": 0.4384,
|
| 2809 |
+
"step": 3680
|
| 2810 |
+
},
|
| 2811 |
+
{
|
| 2812 |
+
"epoch": 1.7291471415182755,
|
| 2813 |
+
"grad_norm": 3.0865719318389893,
|
| 2814 |
+
"learning_rate": 8.919283036551079e-06,
|
| 2815 |
+
"loss": 0.4248,
|
| 2816 |
+
"step": 3690
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 1.7338331771321462,
|
| 2820 |
+
"grad_norm": 3.9282147884368896,
|
| 2821 |
+
"learning_rate": 8.91635426429241e-06,
|
| 2822 |
+
"loss": 0.4231,
|
| 2823 |
+
"step": 3700
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 1.738519212746017,
|
| 2827 |
+
"grad_norm": 3.9746885299682617,
|
| 2828 |
+
"learning_rate": 8.91342549203374e-06,
|
| 2829 |
+
"loss": 0.4152,
|
| 2830 |
+
"step": 3710
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 1.7432052483598874,
|
| 2834 |
+
"grad_norm": 3.8340625762939453,
|
| 2835 |
+
"learning_rate": 8.910496719775071e-06,
|
| 2836 |
+
"loss": 0.4458,
|
| 2837 |
+
"step": 3720
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 1.7478912839737581,
|
| 2841 |
+
"grad_norm": 4.861859321594238,
|
| 2842 |
+
"learning_rate": 8.907567947516402e-06,
|
| 2843 |
+
"loss": 0.4274,
|
| 2844 |
+
"step": 3730
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 1.7525773195876289,
|
| 2848 |
+
"grad_norm": 3.3457283973693848,
|
| 2849 |
+
"learning_rate": 8.904639175257732e-06,
|
| 2850 |
+
"loss": 0.4534,
|
| 2851 |
+
"step": 3740
|
| 2852 |
+
},
|
| 2853 |
+
{
|
| 2854 |
+
"epoch": 1.7572633552014996,
|
| 2855 |
+
"grad_norm": 4.057953834533691,
|
| 2856 |
+
"learning_rate": 8.901710402999063e-06,
|
| 2857 |
+
"loss": 0.484,
|
| 2858 |
+
"step": 3750
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 1.7572633552014996,
|
| 2862 |
+
"eval_loss": 0.05240313336253166,
|
| 2863 |
+
"eval_pearson_cosine": 0.7879299521989642,
|
| 2864 |
+
"eval_pearson_dot": 0.6605985065084816,
|
| 2865 |
+
"eval_pearson_euclidean": 0.7797438530556207,
|
| 2866 |
+
"eval_pearson_manhattan": 0.778216782480726,
|
| 2867 |
+
"eval_runtime": 44.9916,
|
| 2868 |
+
"eval_samples_per_second": 33.34,
|
| 2869 |
+
"eval_spearman_cosine": 0.7888982276270184,
|
| 2870 |
+
"eval_spearman_dot": 0.6669965792210436,
|
| 2871 |
+
"eval_spearman_euclidean": 0.7899037728263932,
|
| 2872 |
+
"eval_spearman_manhattan": 0.7886320032383264,
|
| 2873 |
+
"eval_steps_per_second": 33.34,
|
| 2874 |
+
"step": 3750
|
| 2875 |
+
},
|
| 2876 |
+
{
|
| 2877 |
+
"epoch": 1.7619493908153703,
|
| 2878 |
+
"grad_norm": 3.281102418899536,
|
| 2879 |
+
"learning_rate": 8.898781630740394e-06,
|
| 2880 |
+
"loss": 0.4074,
|
| 2881 |
+
"step": 3760
|
| 2882 |
+
},
|
| 2883 |
+
{
|
| 2884 |
+
"epoch": 1.7666354264292408,
|
| 2885 |
+
"grad_norm": 4.710203170776367,
|
| 2886 |
+
"learning_rate": 8.895852858481725e-06,
|
| 2887 |
+
"loss": 0.4537,
|
| 2888 |
+
"step": 3770
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 1.7713214620431117,
|
| 2892 |
+
"grad_norm": 4.636346817016602,
|
| 2893 |
+
"learning_rate": 8.892924086223056e-06,
|
| 2894 |
+
"loss": 0.4348,
|
| 2895 |
+
"step": 3780
|
| 2896 |
+
},
|
| 2897 |
+
{
|
| 2898 |
+
"epoch": 1.7760074976569822,
|
| 2899 |
+
"grad_norm": 4.518571376800537,
|
| 2900 |
+
"learning_rate": 8.889995313964388e-06,
|
| 2901 |
+
"loss": 0.4515,
|
| 2902 |
+
"step": 3790
|
| 2903 |
+
},
|
| 2904 |
+
{
|
| 2905 |
+
"epoch": 1.780693533270853,
|
| 2906 |
+
"grad_norm": 4.0576066970825195,
|
| 2907 |
+
"learning_rate": 8.887066541705717e-06,
|
| 2908 |
+
"loss": 0.4276,
|
| 2909 |
+
"step": 3800
|
| 2910 |
+
},
|
| 2911 |
+
{
|
| 2912 |
+
"epoch": 1.7853795688847236,
|
| 2913 |
+
"grad_norm": 5.657445430755615,
|
| 2914 |
+
"learning_rate": 8.884137769447048e-06,
|
| 2915 |
+
"loss": 0.4277,
|
| 2916 |
+
"step": 3810
|
| 2917 |
+
},
|
| 2918 |
+
{
|
| 2919 |
+
"epoch": 1.790065604498594,
|
| 2920 |
+
"grad_norm": 5.393405437469482,
|
| 2921 |
+
"learning_rate": 8.88120899718838e-06,
|
| 2922 |
+
"loss": 0.428,
|
| 2923 |
+
"step": 3820
|
| 2924 |
+
},
|
| 2925 |
+
{
|
| 2926 |
+
"epoch": 1.794751640112465,
|
| 2927 |
+
"grad_norm": 4.101112365722656,
|
| 2928 |
+
"learning_rate": 8.87828022492971e-06,
|
| 2929 |
+
"loss": 0.4489,
|
| 2930 |
+
"step": 3830
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 1.7994376757263355,
|
| 2934 |
+
"grad_norm": 3.531888246536255,
|
| 2935 |
+
"learning_rate": 8.87535145267104e-06,
|
| 2936 |
+
"loss": 0.3673,
|
| 2937 |
+
"step": 3840
|
| 2938 |
+
},
|
| 2939 |
+
{
|
| 2940 |
+
"epoch": 1.8041237113402062,
|
| 2941 |
+
"grad_norm": 3.4490315914154053,
|
| 2942 |
+
"learning_rate": 8.872422680412371e-06,
|
| 2943 |
+
"loss": 0.4059,
|
| 2944 |
+
"step": 3850
|
| 2945 |
+
},
|
| 2946 |
+
{
|
| 2947 |
+
"epoch": 1.808809746954077,
|
| 2948 |
+
"grad_norm": 3.034252643585205,
|
| 2949 |
+
"learning_rate": 8.869493908153702e-06,
|
| 2950 |
+
"loss": 0.3832,
|
| 2951 |
+
"step": 3860
|
| 2952 |
+
},
|
| 2953 |
+
{
|
| 2954 |
+
"epoch": 1.8134957825679474,
|
| 2955 |
+
"grad_norm": 4.064283847808838,
|
| 2956 |
+
"learning_rate": 8.866565135895034e-06,
|
| 2957 |
+
"loss": 0.4704,
|
| 2958 |
+
"step": 3870
|
| 2959 |
+
},
|
| 2960 |
+
{
|
| 2961 |
+
"epoch": 1.8181818181818183,
|
| 2962 |
+
"grad_norm": 3.2689194679260254,
|
| 2963 |
+
"learning_rate": 8.863636363636365e-06,
|
| 2964 |
+
"loss": 0.4428,
|
| 2965 |
+
"step": 3880
|
| 2966 |
+
},
|
| 2967 |
+
{
|
| 2968 |
+
"epoch": 1.8228678537956888,
|
| 2969 |
+
"grad_norm": 3.173530101776123,
|
| 2970 |
+
"learning_rate": 8.860707591377694e-06,
|
| 2971 |
+
"loss": 0.4283,
|
| 2972 |
+
"step": 3890
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 1.8275538894095595,
|
| 2976 |
+
"grad_norm": 3.638122081756592,
|
| 2977 |
+
"learning_rate": 8.857778819119026e-06,
|
| 2978 |
+
"loss": 0.4225,
|
| 2979 |
+
"step": 3900
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 1.8322399250234302,
|
| 2983 |
+
"grad_norm": 3.636679172515869,
|
| 2984 |
+
"learning_rate": 8.854850046860357e-06,
|
| 2985 |
+
"loss": 0.4154,
|
| 2986 |
+
"step": 3910
|
| 2987 |
+
},
|
| 2988 |
+
{
|
| 2989 |
+
"epoch": 1.8369259606373007,
|
| 2990 |
+
"grad_norm": 3.810847520828247,
|
| 2991 |
+
"learning_rate": 8.851921274601688e-06,
|
| 2992 |
+
"loss": 0.3931,
|
| 2993 |
+
"step": 3920
|
| 2994 |
+
},
|
| 2995 |
+
{
|
| 2996 |
+
"epoch": 1.8416119962511717,
|
| 2997 |
+
"grad_norm": 3.7469394207000732,
|
| 2998 |
+
"learning_rate": 8.848992502343019e-06,
|
| 2999 |
+
"loss": 0.4472,
|
| 3000 |
+
"step": 3930
|
| 3001 |
+
},
|
| 3002 |
+
{
|
| 3003 |
+
"epoch": 1.8462980318650422,
|
| 3004 |
+
"grad_norm": 4.962492942810059,
|
| 3005 |
+
"learning_rate": 8.846063730084349e-06,
|
| 3006 |
+
"loss": 0.4324,
|
| 3007 |
+
"step": 3940
|
| 3008 |
+
},
|
| 3009 |
+
{
|
| 3010 |
+
"epoch": 1.8509840674789129,
|
| 3011 |
+
"grad_norm": 3.4641172885894775,
|
| 3012 |
+
"learning_rate": 8.84313495782568e-06,
|
| 3013 |
+
"loss": 0.4234,
|
| 3014 |
+
"step": 3950
|
| 3015 |
+
},
|
| 3016 |
+
{
|
| 3017 |
+
"epoch": 1.8556701030927836,
|
| 3018 |
+
"grad_norm": 3.8601555824279785,
|
| 3019 |
+
"learning_rate": 8.840206185567011e-06,
|
| 3020 |
+
"loss": 0.4045,
|
| 3021 |
+
"step": 3960
|
| 3022 |
+
},
|
| 3023 |
+
{
|
| 3024 |
+
"epoch": 1.860356138706654,
|
| 3025 |
+
"grad_norm": 6.290759086608887,
|
| 3026 |
+
"learning_rate": 8.837277413308342e-06,
|
| 3027 |
+
"loss": 0.4655,
|
| 3028 |
+
"step": 3970
|
| 3029 |
+
},
|
| 3030 |
+
{
|
| 3031 |
+
"epoch": 1.865042174320525,
|
| 3032 |
+
"grad_norm": 3.5882256031036377,
|
| 3033 |
+
"learning_rate": 8.834348641049673e-06,
|
| 3034 |
+
"loss": 0.4298,
|
| 3035 |
+
"step": 3980
|
| 3036 |
+
},
|
| 3037 |
+
{
|
| 3038 |
+
"epoch": 1.8697282099343955,
|
| 3039 |
+
"grad_norm": 3.133535623550415,
|
| 3040 |
+
"learning_rate": 8.831419868791003e-06,
|
| 3041 |
+
"loss": 0.4508,
|
| 3042 |
+
"step": 3990
|
| 3043 |
+
},
|
| 3044 |
+
{
|
| 3045 |
+
"epoch": 1.8744142455482662,
|
| 3046 |
+
"grad_norm": 3.220383644104004,
|
| 3047 |
+
"learning_rate": 8.828491096532334e-06,
|
| 3048 |
+
"loss": 0.4348,
|
| 3049 |
+
"step": 4000
|
| 3050 |
+
},
|
| 3051 |
+
{
|
| 3052 |
+
"epoch": 1.8744142455482662,
|
| 3053 |
+
"eval_loss": 0.04981923848390579,
|
| 3054 |
+
"eval_pearson_cosine": 0.790612878761543,
|
| 3055 |
+
"eval_pearson_dot": 0.6612786229229286,
|
| 3056 |
+
"eval_pearson_euclidean": 0.7799249806775554,
|
| 3057 |
+
"eval_pearson_manhattan": 0.7784476870813819,
|
| 3058 |
+
"eval_runtime": 45.9371,
|
| 3059 |
+
"eval_samples_per_second": 32.653,
|
| 3060 |
+
"eval_spearman_cosine": 0.7908100570922554,
|
| 3061 |
+
"eval_spearman_dot": 0.6689224987064551,
|
| 3062 |
+
"eval_spearman_euclidean": 0.7902520878335856,
|
| 3063 |
+
"eval_spearman_manhattan": 0.7892503488739743,
|
| 3064 |
+
"eval_steps_per_second": 32.653,
|
| 3065 |
+
"step": 4000
|
| 3066 |
}
|
| 3067 |
],
|
| 3068 |
"logging_steps": 10,
|