Training in progress, epoch 0, checkpoint
Browse files- last-checkpoint/adapter_model.safetensors +1 -1
- last-checkpoint/global_step200/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step200/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step200/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step200/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step200/mp_rank_00_model_states.pt +3 -0
- last-checkpoint/latest +1 -1
- last-checkpoint/rng_state_0.pth +1 -1
- last-checkpoint/rng_state_1.pth +1 -1
- last-checkpoint/rng_state_2.pth +1 -1
- last-checkpoint/rng_state_3.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +501 -3
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 108701128
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3ec6f7626957aef3ff168ab3aa0454634adf31bc9aae7d2893913597fac45def
|
| 3 |
size 108701128
|
last-checkpoint/global_step200/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:72fb69ed897b2f971f53a25dc45b3b374c75c6edd9f83a337e1a4c7efe3d4f23
|
| 3 |
+
size 81943013
|
last-checkpoint/global_step200/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:839efbb9b3b68ab48b93db19e5da9923a51a10c4e820f8404061894ade8c0bdc
|
| 3 |
+
size 81943013
|
last-checkpoint/global_step200/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ad0e0b0a5b036a2866633579144fbbc26ed1bf78bdb027e1c958786ba04bb0e5
|
| 3 |
+
size 81943077
|
last-checkpoint/global_step200/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ba248ce5963316b7fb840cadc54c0241b06b2fb71baa3c07e3fb7e977cf20316
|
| 3 |
+
size 81943077
|
last-checkpoint/global_step200/mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f0b31d9e20161f24ef233e81a6b9e00ed8b9b6a5d1a44b4f53e1e56905756368
|
| 3 |
+
size 381105765
|
last-checkpoint/latest
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
|
|
|
|
| 1 |
+
global_step200
|
last-checkpoint/rng_state_0.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15429
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:81d5f83aeb4b3f559bd28377336d47659b320e7f6ef2e5a723d284716278a151
|
| 3 |
size 15429
|
last-checkpoint/rng_state_1.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15429
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2626437dcb133ffcf003ac89603f8cce07459b93a98d760cd9419e0d6a994067
|
| 3 |
size 15429
|
last-checkpoint/rng_state_2.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15429
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ae777e24d50cb7159634e1245f0697ba0fc64d5b26d535f2c80e411371a90b1c
|
| 3 |
size 15429
|
last-checkpoint/rng_state_3.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15429
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:afc5a67564eebcfc961e8f1406a7418cc73497c2935a39af0232ef59f8153a6a
|
| 3 |
size 15429
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1401
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0c8a0f1df51a7687aaa6cccd4e16264648b2924176d968dccc7e021f62f9ad6f
|
| 3 |
size 1401
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
"best_global_step": null,
|
| 3 |
-
"best_metric": 0.
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 50,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -174,6 +174,504 @@
|
|
| 174 |
"eval_samples_per_second": 118.251,
|
| 175 |
"eval_steps_per_second": 14.781,
|
| 176 |
"step": 50
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 177 |
}
|
| 178 |
],
|
| 179 |
"logging_steps": 5,
|
|
|
|
| 1 |
{
|
| 2 |
"best_global_step": null,
|
| 3 |
+
"best_metric": 0.0017637914279475808,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.24464831804281345,
|
| 6 |
"eval_steps": 50,
|
| 7 |
+
"global_step": 200,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 174 |
"eval_samples_per_second": 118.251,
|
| 175 |
"eval_steps_per_second": 14.781,
|
| 176 |
"step": 50
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.0672782874617737,
|
| 180 |
+
"grad_norm": 0.001428966992534697,
|
| 181 |
+
"learning_rate": 4.9781466138752906e-05,
|
| 182 |
+
"logits/chosen": -1.228124976158142,
|
| 183 |
+
"logits/rejected": NaN,
|
| 184 |
+
"logps/chosen": -262.54998779296875,
|
| 185 |
+
"logps/rejected": -423.79998779296875,
|
| 186 |
+
"loss": 0.0173,
|
| 187 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 188 |
+
"rewards/chosen": -2.325390577316284,
|
| 189 |
+
"rewards/margins": 18.5,
|
| 190 |
+
"rewards/rejected": -20.825000762939453,
|
| 191 |
+
"step": 55
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.07339449541284404,
|
| 195 |
+
"grad_norm": 3.8112677430035546e-05,
|
| 196 |
+
"learning_rate": 5.4390861151600386e-05,
|
| 197 |
+
"logits/chosen": -1.1671874523162842,
|
| 198 |
+
"logits/rejected": -0.6392577886581421,
|
| 199 |
+
"logps/chosen": -239.1999969482422,
|
| 200 |
+
"logps/rejected": -433.3999938964844,
|
| 201 |
+
"loss": 0.0,
|
| 202 |
+
"rewards/accuracies": 1.0,
|
| 203 |
+
"rewards/chosen": -3.4078125953674316,
|
| 204 |
+
"rewards/margins": 18.274999618530273,
|
| 205 |
+
"rewards/rejected": -21.662500381469727,
|
| 206 |
+
"step": 60
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.07951070336391437,
|
| 210 |
+
"grad_norm": 5.469312600325793e-05,
|
| 211 |
+
"learning_rate": 5.900025616444788e-05,
|
| 212 |
+
"logits/chosen": -1.118749976158142,
|
| 213 |
+
"logits/rejected": NaN,
|
| 214 |
+
"logps/chosen": -244.5,
|
| 215 |
+
"logps/rejected": -417.0,
|
| 216 |
+
"loss": 0.0086,
|
| 217 |
+
"rewards/accuracies": 0.987500011920929,
|
| 218 |
+
"rewards/chosen": -2.7113280296325684,
|
| 219 |
+
"rewards/margins": 18.700000762939453,
|
| 220 |
+
"rewards/rejected": -21.412500381469727,
|
| 221 |
+
"step": 65
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.0856269113149847,
|
| 225 |
+
"grad_norm": 1.760563100106083e-05,
|
| 226 |
+
"learning_rate": 6.360965117729536e-05,
|
| 227 |
+
"logits/chosen": -0.9232177734375,
|
| 228 |
+
"logits/rejected": -0.4688720703125,
|
| 229 |
+
"logps/chosen": -275.20001220703125,
|
| 230 |
+
"logps/rejected": -442.0,
|
| 231 |
+
"loss": 0.0,
|
| 232 |
+
"rewards/accuracies": 1.0,
|
| 233 |
+
"rewards/chosen": -3.489062547683716,
|
| 234 |
+
"rewards/margins": 18.725000381469727,
|
| 235 |
+
"rewards/rejected": -22.212499618530273,
|
| 236 |
+
"step": 70
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.09174311926605505,
|
| 240 |
+
"grad_norm": 58.193965911865234,
|
| 241 |
+
"learning_rate": 6.821904619014286e-05,
|
| 242 |
+
"logits/chosen": -0.9205077886581421,
|
| 243 |
+
"logits/rejected": NaN,
|
| 244 |
+
"logps/chosen": -290.29998779296875,
|
| 245 |
+
"logps/rejected": -420.20001220703125,
|
| 246 |
+
"loss": 0.0536,
|
| 247 |
+
"rewards/accuracies": 0.949999988079071,
|
| 248 |
+
"rewards/chosen": -2.6796875,
|
| 249 |
+
"rewards/margins": 17.637500762939453,
|
| 250 |
+
"rewards/rejected": -20.337499618530273,
|
| 251 |
+
"step": 75
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.09785932721712538,
|
| 255 |
+
"grad_norm": 2.4334192403330235e-06,
|
| 256 |
+
"learning_rate": 7.282844120299036e-05,
|
| 257 |
+
"logits/chosen": -0.6377929449081421,
|
| 258 |
+
"logits/rejected": NaN,
|
| 259 |
+
"logps/chosen": -359.3999938964844,
|
| 260 |
+
"logps/rejected": -455.3999938964844,
|
| 261 |
+
"loss": 0.201,
|
| 262 |
+
"rewards/accuracies": 0.9375,
|
| 263 |
+
"rewards/chosen": -2.579296827316284,
|
| 264 |
+
"rewards/margins": 21.993749618530273,
|
| 265 |
+
"rewards/rejected": -24.575000762939453,
|
| 266 |
+
"step": 80
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.10397553516819572,
|
| 270 |
+
"grad_norm": 3.0753548579259826e-12,
|
| 271 |
+
"learning_rate": 7.743783621583784e-05,
|
| 272 |
+
"logits/chosen": -1.0593750476837158,
|
| 273 |
+
"logits/rejected": -0.25745850801467896,
|
| 274 |
+
"logps/chosen": -228.5500030517578,
|
| 275 |
+
"logps/rejected": -593.4000244140625,
|
| 276 |
+
"loss": 0.0017,
|
| 277 |
+
"rewards/accuracies": 1.0,
|
| 278 |
+
"rewards/chosen": -5.3359375,
|
| 279 |
+
"rewards/margins": 31.037500381469727,
|
| 280 |
+
"rewards/rejected": -36.38750076293945,
|
| 281 |
+
"step": 85
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.11009174311926606,
|
| 285 |
+
"grad_norm": 1.4482483268185575e-10,
|
| 286 |
+
"learning_rate": 8.204723122868534e-05,
|
| 287 |
+
"logits/chosen": -0.49467772245407104,
|
| 288 |
+
"logits/rejected": 0.0545654296875,
|
| 289 |
+
"logps/chosen": -405.79998779296875,
|
| 290 |
+
"logps/rejected": -918.4000244140625,
|
| 291 |
+
"loss": 0.0,
|
| 292 |
+
"rewards/accuracies": 1.0,
|
| 293 |
+
"rewards/chosen": -16.287500381469727,
|
| 294 |
+
"rewards/margins": 52.0,
|
| 295 |
+
"rewards/rejected": -68.19999694824219,
|
| 296 |
+
"step": 90
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.1162079510703364,
|
| 300 |
+
"grad_norm": 1.7419449400719184e-14,
|
| 301 |
+
"learning_rate": 8.665662624153282e-05,
|
| 302 |
+
"logits/chosen": -1.132421851158142,
|
| 303 |
+
"logits/rejected": NaN,
|
| 304 |
+
"logps/chosen": -331.70001220703125,
|
| 305 |
+
"logps/rejected": -948.4000244140625,
|
| 306 |
+
"loss": 0.4562,
|
| 307 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 308 |
+
"rewards/chosen": -11.300000190734863,
|
| 309 |
+
"rewards/margins": 62.125,
|
| 310 |
+
"rewards/rejected": -73.4749984741211,
|
| 311 |
+
"step": 95
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.12232415902140673,
|
| 315 |
+
"grad_norm": 147.80014038085938,
|
| 316 |
+
"learning_rate": 9.126602125438031e-05,
|
| 317 |
+
"logits/chosen": -1.4375,
|
| 318 |
+
"logits/rejected": -1.0437500476837158,
|
| 319 |
+
"logps/chosen": -227.39999389648438,
|
| 320 |
+
"logps/rejected": -900.4000244140625,
|
| 321 |
+
"loss": 0.5125,
|
| 322 |
+
"rewards/accuracies": 0.987500011920929,
|
| 323 |
+
"rewards/chosen": -6.349999904632568,
|
| 324 |
+
"rewards/margins": 58.875,
|
| 325 |
+
"rewards/rejected": -65.2249984741211,
|
| 326 |
+
"step": 100
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.12232415902140673,
|
| 330 |
+
"eval_logits/chosen": -1.3658323287963867,
|
| 331 |
+
"eval_logits/rejected": NaN,
|
| 332 |
+
"eval_logps/chosen": -275.61224365234375,
|
| 333 |
+
"eval_logps/rejected": -1236.89794921875,
|
| 334 |
+
"eval_loss": 0.21779336035251617,
|
| 335 |
+
"eval_rewards/accuracies": 0.9923469424247742,
|
| 336 |
+
"eval_rewards/chosen": -7.099011421203613,
|
| 337 |
+
"eval_rewards/margins": 91.91836547851562,
|
| 338 |
+
"eval_rewards/rejected": -99.04081726074219,
|
| 339 |
+
"eval_runtime": 3.1052,
|
| 340 |
+
"eval_samples_per_second": 126.241,
|
| 341 |
+
"eval_steps_per_second": 15.78,
|
| 342 |
+
"step": 100
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 0.12844036697247707,
|
| 346 |
+
"grad_norm": 9.682947977770899e-21,
|
| 347 |
+
"learning_rate": 9.218786312367835e-05,
|
| 348 |
+
"logits/chosen": -1.560156226158142,
|
| 349 |
+
"logits/rejected": -1.505468726158142,
|
| 350 |
+
"logps/chosen": -322.8999938964844,
|
| 351 |
+
"logps/rejected": -1656.800048828125,
|
| 352 |
+
"loss": 0.0,
|
| 353 |
+
"rewards/accuracies": 1.0,
|
| 354 |
+
"rewards/chosen": -13.934374809265137,
|
| 355 |
+
"rewards/margins": 126.9000015258789,
|
| 356 |
+
"rewards/rejected": -140.89999389648438,
|
| 357 |
+
"step": 105
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.1345565749235474,
|
| 361 |
+
"grad_norm": 3.848379052308646e-17,
|
| 362 |
+
"learning_rate": 9.21877122698656e-05,
|
| 363 |
+
"logits/chosen": -1.169531226158142,
|
| 364 |
+
"logits/rejected": -1.3546874523162842,
|
| 365 |
+
"logps/chosen": -569.4000244140625,
|
| 366 |
+
"logps/rejected": -1708.0,
|
| 367 |
+
"loss": 4.1156,
|
| 368 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 369 |
+
"rewards/chosen": -33.493751525878906,
|
| 370 |
+
"rewards/margins": 111.4749984741211,
|
| 371 |
+
"rewards/rejected": -144.9499969482422,
|
| 372 |
+
"step": 110
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 0.14067278287461774,
|
| 376 |
+
"grad_norm": 3.294127637332167e-09,
|
| 377 |
+
"learning_rate": 9.218744537506154e-05,
|
| 378 |
+
"logits/chosen": -0.18134765326976776,
|
| 379 |
+
"logits/rejected": -0.7176758050918579,
|
| 380 |
+
"logps/chosen": -281.3999938964844,
|
| 381 |
+
"logps/rejected": -832.4000244140625,
|
| 382 |
+
"loss": 0.0,
|
| 383 |
+
"rewards/accuracies": 1.0,
|
| 384 |
+
"rewards/chosen": -7.678124904632568,
|
| 385 |
+
"rewards/margins": 51.17499923706055,
|
| 386 |
+
"rewards/rejected": -58.900001525878906,
|
| 387 |
+
"step": 115
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.14678899082568808,
|
| 391 |
+
"grad_norm": 0.2668285369873047,
|
| 392 |
+
"learning_rate": 9.21870624399381e-05,
|
| 393 |
+
"logits/chosen": 0.09755859524011612,
|
| 394 |
+
"logits/rejected": -0.16208496689796448,
|
| 395 |
+
"logps/chosen": -317.29998779296875,
|
| 396 |
+
"logps/rejected": -565.5999755859375,
|
| 397 |
+
"loss": 0.0,
|
| 398 |
+
"rewards/accuracies": 1.0,
|
| 399 |
+
"rewards/chosen": -8.418749809265137,
|
| 400 |
+
"rewards/margins": 25.875,
|
| 401 |
+
"rewards/rejected": -34.275001525878906,
|
| 402 |
+
"step": 120
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.1529051987767584,
|
| 406 |
+
"grad_norm": 142.54164123535156,
|
| 407 |
+
"learning_rate": 9.218656346545931e-05,
|
| 408 |
+
"logits/chosen": 0.3714355528354645,
|
| 409 |
+
"logits/rejected": NaN,
|
| 410 |
+
"logps/chosen": -339.1000061035156,
|
| 411 |
+
"logps/rejected": -609.7999877929688,
|
| 412 |
+
"loss": 0.6578,
|
| 413 |
+
"rewards/accuracies": 0.949999988079071,
|
| 414 |
+
"rewards/chosen": -9.662500381469727,
|
| 415 |
+
"rewards/margins": 28.75,
|
| 416 |
+
"rewards/rejected": -38.45000076293945,
|
| 417 |
+
"step": 125
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.15902140672782875,
|
| 421 |
+
"grad_norm": 0.0030178045853972435,
|
| 422 |
+
"learning_rate": 9.218594845288136e-05,
|
| 423 |
+
"logits/chosen": 0.08590088039636612,
|
| 424 |
+
"logits/rejected": 0.2857910096645355,
|
| 425 |
+
"logps/chosen": -277.1000061035156,
|
| 426 |
+
"logps/rejected": -548.7999877929688,
|
| 427 |
+
"loss": 0.0,
|
| 428 |
+
"rewards/accuracies": 1.0,
|
| 429 |
+
"rewards/chosen": -10.225000381469727,
|
| 430 |
+
"rewards/margins": 22.0,
|
| 431 |
+
"rewards/rejected": -32.224998474121094,
|
| 432 |
+
"step": 130
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 0.1651376146788991,
|
| 436 |
+
"grad_norm": 55.50654220581055,
|
| 437 |
+
"learning_rate": 9.218521740375251e-05,
|
| 438 |
+
"logits/chosen": -0.46796876192092896,
|
| 439 |
+
"logits/rejected": NaN,
|
| 440 |
+
"logps/chosen": -419.79998779296875,
|
| 441 |
+
"logps/rejected": -670.0,
|
| 442 |
+
"loss": 0.1262,
|
| 443 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 444 |
+
"rewards/chosen": -10.399999618530273,
|
| 445 |
+
"rewards/margins": 35.462501525878906,
|
| 446 |
+
"rewards/rejected": -45.88750076293945,
|
| 447 |
+
"step": 135
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.1712538226299694,
|
| 451 |
+
"grad_norm": 70.77788543701172,
|
| 452 |
+
"learning_rate": 9.21843703199132e-05,
|
| 453 |
+
"logits/chosen": -1.024999976158142,
|
| 454 |
+
"logits/rejected": -0.693359375,
|
| 455 |
+
"logps/chosen": -475.3999938964844,
|
| 456 |
+
"logps/rejected": -902.0,
|
| 457 |
+
"loss": 0.3516,
|
| 458 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 459 |
+
"rewards/chosen": -23.200000762939453,
|
| 460 |
+
"rewards/margins": 45.525001525878906,
|
| 461 |
+
"rewards/rejected": -68.69999694824219,
|
| 462 |
+
"step": 140
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 0.17737003058103976,
|
| 466 |
+
"grad_norm": 3.293635381851345e-05,
|
| 467 |
+
"learning_rate": 9.218340720349596e-05,
|
| 468 |
+
"logits/chosen": -0.7054687738418579,
|
| 469 |
+
"logits/rejected": -0.32458192110061646,
|
| 470 |
+
"logps/chosen": -446.3999938964844,
|
| 471 |
+
"logps/rejected": -905.5999755859375,
|
| 472 |
+
"loss": 0.0,
|
| 473 |
+
"rewards/accuracies": 1.0,
|
| 474 |
+
"rewards/chosen": -22.450000762939453,
|
| 475 |
+
"rewards/margins": 43.724998474121094,
|
| 476 |
+
"rewards/rejected": -66.1500015258789,
|
| 477 |
+
"step": 145
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.1834862385321101,
|
| 481 |
+
"grad_norm": 4.300224816233822e-07,
|
| 482 |
+
"learning_rate": 9.218232805692544e-05,
|
| 483 |
+
"logits/chosen": -0.634448230266571,
|
| 484 |
+
"logits/rejected": -0.5257812738418579,
|
| 485 |
+
"logps/chosen": -407.1000061035156,
|
| 486 |
+
"logps/rejected": -775.5999755859375,
|
| 487 |
+
"loss": 0.0,
|
| 488 |
+
"rewards/accuracies": 1.0,
|
| 489 |
+
"rewards/chosen": -19.387500762939453,
|
| 490 |
+
"rewards/margins": 35.037498474121094,
|
| 491 |
+
"rewards/rejected": -54.45000076293945,
|
| 492 |
+
"step": 150
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.1834862385321101,
|
| 496 |
+
"eval_logits/chosen": -0.8233518004417419,
|
| 497 |
+
"eval_logits/rejected": NaN,
|
| 498 |
+
"eval_logps/chosen": -414.73468017578125,
|
| 499 |
+
"eval_logps/rejected": -896.9796142578125,
|
| 500 |
+
"eval_loss": 0.11673443019390106,
|
| 501 |
+
"eval_rewards/accuracies": 0.9948979616165161,
|
| 502 |
+
"eval_rewards/chosen": -20.599489212036133,
|
| 503 |
+
"eval_rewards/margins": 45.46938705444336,
|
| 504 |
+
"eval_rewards/rejected": -66.08163452148438,
|
| 505 |
+
"eval_runtime": 3.1768,
|
| 506 |
+
"eval_samples_per_second": 123.395,
|
| 507 |
+
"eval_steps_per_second": 15.424,
|
| 508 |
+
"step": 150
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.18960244648318042,
|
| 512 |
+
"grad_norm": 0.0,
|
| 513 |
+
"learning_rate": 9.218113288291835e-05,
|
| 514 |
+
"logits/chosen": -0.681640625,
|
| 515 |
+
"logits/rejected": NaN,
|
| 516 |
+
"logps/chosen": -397.79998779296875,
|
| 517 |
+
"logps/rejected": -1483.199951171875,
|
| 518 |
+
"loss": 0.6354,
|
| 519 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 520 |
+
"rewards/chosen": -13.587499618530273,
|
| 521 |
+
"rewards/margins": 110.57499694824219,
|
| 522 |
+
"rewards/rejected": -124.0250015258789,
|
| 523 |
+
"step": 155
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 0.19571865443425077,
|
| 527 |
+
"grad_norm": 0.0,
|
| 528 |
+
"learning_rate": 9.217982168448358e-05,
|
| 529 |
+
"logits/chosen": -0.910449206829071,
|
| 530 |
+
"logits/rejected": NaN,
|
| 531 |
+
"logps/chosen": -342.8999938964844,
|
| 532 |
+
"logps/rejected": -1642.4000244140625,
|
| 533 |
+
"loss": 0.0391,
|
| 534 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 535 |
+
"rewards/chosen": -6.589062690734863,
|
| 536 |
+
"rewards/margins": 135.10000610351562,
|
| 537 |
+
"rewards/rejected": -141.60000610351562,
|
| 538 |
+
"step": 160
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.2018348623853211,
|
| 542 |
+
"grad_norm": 70.26122283935547,
|
| 543 |
+
"learning_rate": 9.217839446492207e-05,
|
| 544 |
+
"logits/chosen": -0.760546863079071,
|
| 545 |
+
"logits/rejected": NaN,
|
| 546 |
+
"logps/chosen": -298.5,
|
| 547 |
+
"logps/rejected": -1636.0,
|
| 548 |
+
"loss": 0.2031,
|
| 549 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 550 |
+
"rewards/chosen": -7.454687595367432,
|
| 551 |
+
"rewards/margins": 130.1999969482422,
|
| 552 |
+
"rewards/rejected": -137.6999969482422,
|
| 553 |
+
"step": 165
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 0.20795107033639143,
|
| 557 |
+
"grad_norm": 0.0,
|
| 558 |
+
"learning_rate": 9.217685122782683e-05,
|
| 559 |
+
"logits/chosen": -0.618701159954071,
|
| 560 |
+
"logits/rejected": -1.9210937023162842,
|
| 561 |
+
"logps/chosen": -258.8999938964844,
|
| 562 |
+
"logps/rejected": -1873.5999755859375,
|
| 563 |
+
"loss": 0.0,
|
| 564 |
+
"rewards/accuracies": 1.0,
|
| 565 |
+
"rewards/chosen": -5.925000190734863,
|
| 566 |
+
"rewards/margins": 153.35000610351562,
|
| 567 |
+
"rewards/rejected": -159.3000030517578,
|
| 568 |
+
"step": 170
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 0.21406727828746178,
|
| 572 |
+
"grad_norm": 0.0,
|
| 573 |
+
"learning_rate": 9.217519197708296e-05,
|
| 574 |
+
"logits/chosen": -0.6034179925918579,
|
| 575 |
+
"logits/rejected": -2.01171875,
|
| 576 |
+
"logps/chosen": -321.3999938964844,
|
| 577 |
+
"logps/rejected": -1909.5999755859375,
|
| 578 |
+
"loss": 0.3109,
|
| 579 |
+
"rewards/accuracies": 0.987500011920929,
|
| 580 |
+
"rewards/chosen": -5.234375,
|
| 581 |
+
"rewards/margins": 157.5500030517578,
|
| 582 |
+
"rewards/rejected": -163.0,
|
| 583 |
+
"step": 175
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.22018348623853212,
|
| 587 |
+
"grad_norm": 0.0,
|
| 588 |
+
"learning_rate": 9.21734167168676e-05,
|
| 589 |
+
"logits/chosen": -0.517285168170929,
|
| 590 |
+
"logits/rejected": NaN,
|
| 591 |
+
"logps/chosen": -339.0,
|
| 592 |
+
"logps/rejected": -2012.800048828125,
|
| 593 |
+
"loss": 0.2031,
|
| 594 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 595 |
+
"rewards/chosen": -7.631249904632568,
|
| 596 |
+
"rewards/margins": 166.1999969482422,
|
| 597 |
+
"rewards/rejected": -173.89999389648438,
|
| 598 |
+
"step": 180
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.22629969418960244,
|
| 602 |
+
"grad_norm": 0.0,
|
| 603 |
+
"learning_rate": 9.217152545165001e-05,
|
| 604 |
+
"logits/chosen": -1.088281273841858,
|
| 605 |
+
"logits/rejected": NaN,
|
| 606 |
+
"logps/chosen": -214.5,
|
| 607 |
+
"logps/rejected": -1929.5999755859375,
|
| 608 |
+
"loss": 0.0087,
|
| 609 |
+
"rewards/accuracies": 0.987500011920929,
|
| 610 |
+
"rewards/chosen": -3.342968702316284,
|
| 611 |
+
"rewards/margins": 164.0500030517578,
|
| 612 |
+
"rewards/rejected": -167.5,
|
| 613 |
+
"step": 185
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 0.2324159021406728,
|
| 617 |
+
"grad_norm": 0.0,
|
| 618 |
+
"learning_rate": 9.216951818619142e-05,
|
| 619 |
+
"logits/chosen": -0.842968761920929,
|
| 620 |
+
"logits/rejected": -2.0843749046325684,
|
| 621 |
+
"logps/chosen": -216.8000030517578,
|
| 622 |
+
"logps/rejected": -2077.60009765625,
|
| 623 |
+
"loss": 0.0,
|
| 624 |
+
"rewards/accuracies": 1.0,
|
| 625 |
+
"rewards/chosen": -1.4921875,
|
| 626 |
+
"rewards/margins": 178.60000610351562,
|
| 627 |
+
"rewards/rejected": -180.10000610351562,
|
| 628 |
+
"step": 190
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 0.23853211009174313,
|
| 632 |
+
"grad_norm": 3.3227390256918218e-19,
|
| 633 |
+
"learning_rate": 9.216739492554512e-05,
|
| 634 |
+
"logits/chosen": -0.9892578125,
|
| 635 |
+
"logits/rejected": NaN,
|
| 636 |
+
"logps/chosen": -266.0,
|
| 637 |
+
"logps/rejected": -1977.5999755859375,
|
| 638 |
+
"loss": 0.0086,
|
| 639 |
+
"rewards/accuracies": 0.987500011920929,
|
| 640 |
+
"rewards/chosen": -1.942968726158142,
|
| 641 |
+
"rewards/margins": 169.4499969482422,
|
| 642 |
+
"rewards/rejected": -171.25,
|
| 643 |
+
"step": 195
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 0.24464831804281345,
|
| 647 |
+
"grad_norm": 0.0,
|
| 648 |
+
"learning_rate": 9.216515567505643e-05,
|
| 649 |
+
"logits/chosen": -1.1339843273162842,
|
| 650 |
+
"logits/rejected": -2.2671875953674316,
|
| 651 |
+
"logps/chosen": -218.4499969482422,
|
| 652 |
+
"logps/rejected": -1971.199951171875,
|
| 653 |
+
"loss": 0.0,
|
| 654 |
+
"rewards/accuracies": 1.0,
|
| 655 |
+
"rewards/chosen": -2.544628858566284,
|
| 656 |
+
"rewards/margins": 168.3000030517578,
|
| 657 |
+
"rewards/rejected": -171.10000610351562,
|
| 658 |
+
"step": 200
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 0.24464831804281345,
|
| 662 |
+
"eval_logits/chosen": -1.1237045526504517,
|
| 663 |
+
"eval_logits/rejected": NaN,
|
| 664 |
+
"eval_logps/chosen": -241.03060913085938,
|
| 665 |
+
"eval_logps/rejected": -2041.142822265625,
|
| 666 |
+
"eval_loss": 0.0017637914279475808,
|
| 667 |
+
"eval_rewards/accuracies": 0.9974489808082581,
|
| 668 |
+
"eval_rewards/chosen": -3.7350003719329834,
|
| 669 |
+
"eval_rewards/margins": 173.2653045654297,
|
| 670 |
+
"eval_rewards/rejected": -176.97959899902344,
|
| 671 |
+
"eval_runtime": 3.16,
|
| 672 |
+
"eval_samples_per_second": 124.052,
|
| 673 |
+
"eval_steps_per_second": 15.507,
|
| 674 |
+
"step": 200
|
| 675 |
}
|
| 676 |
],
|
| 677 |
"logging_steps": 5,
|