Model save
Browse files- README.md +83 -0
- adapter_model.safetensors +1 -1
- all_results.json +21 -0
- eval_results.json +16 -0
- runs/Mar04_11-43-39_gpu4-119-4/events.out.tfevents.1709513183.gpu4-119-4.2293727.0 +2 -2
- runs/Mar04_11-43-39_gpu4-119-4/events.out.tfevents.1709531683.gpu4-119-4.2293727.1 +3 -0
- train_results.json +8 -0
- trainer_state.json +1518 -0
README.md
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
base_model: microsoft/phi-2
|
| 9 |
+
model-index:
|
| 10 |
+
- name: phi-2-dpo-ultrafeedback-lora
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# phi-2-dpo-ultrafeedback-lora
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [microsoft/phi-2](https://huggingface.co/microsoft/phi-2) on the None dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.6537
|
| 22 |
+
- Rewards/chosen: -0.2570
|
| 23 |
+
- Rewards/rejected: -0.3767
|
| 24 |
+
- Rewards/accuracies: 0.6580
|
| 25 |
+
- Rewards/margins: 0.1196
|
| 26 |
+
- Logps/rejected: -269.1014
|
| 27 |
+
- Logps/chosen: -285.9487
|
| 28 |
+
- Logits/rejected: 0.7335
|
| 29 |
+
- Logits/chosen: 0.6309
|
| 30 |
+
|
| 31 |
+
## Model description
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Intended uses & limitations
|
| 36 |
+
|
| 37 |
+
More information needed
|
| 38 |
+
|
| 39 |
+
## Training and evaluation data
|
| 40 |
+
|
| 41 |
+
More information needed
|
| 42 |
+
|
| 43 |
+
## Training procedure
|
| 44 |
+
|
| 45 |
+
### Training hyperparameters
|
| 46 |
+
|
| 47 |
+
The following hyperparameters were used during training:
|
| 48 |
+
- learning_rate: 5e-06
|
| 49 |
+
- train_batch_size: 4
|
| 50 |
+
- eval_batch_size: 4
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- num_devices: 4
|
| 54 |
+
- gradient_accumulation_steps: 4
|
| 55 |
+
- total_train_batch_size: 64
|
| 56 |
+
- total_eval_batch_size: 16
|
| 57 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 58 |
+
- lr_scheduler_type: cosine
|
| 59 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 60 |
+
- num_epochs: 2
|
| 61 |
+
|
| 62 |
+
### Training results
|
| 63 |
+
|
| 64 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 65 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 66 |
+
| 0.6929 | 0.21 | 100 | 0.6928 | 0.0002 | -0.0010 | 0.5320 | 0.0012 | -231.5360 | -260.2240 | 0.9168 | 0.8145 |
|
| 67 |
+
| 0.6893 | 0.42 | 200 | 0.6891 | -0.0038 | -0.0134 | 0.6500 | 0.0096 | -232.7742 | -260.6225 | 0.9234 | 0.8205 |
|
| 68 |
+
| 0.6809 | 0.63 | 300 | 0.6810 | -0.0312 | -0.0611 | 0.6680 | 0.0299 | -237.5431 | -263.3647 | 0.9151 | 0.8092 |
|
| 69 |
+
| 0.6671 | 0.84 | 400 | 0.6723 | -0.0854 | -0.1408 | 0.6640 | 0.0553 | -245.5124 | -268.7867 | 0.8790 | 0.7713 |
|
| 70 |
+
| 0.6627 | 1.05 | 500 | 0.6645 | -0.1494 | -0.2293 | 0.6680 | 0.0799 | -254.3704 | -275.1849 | 0.8294 | 0.7217 |
|
| 71 |
+
| 0.6476 | 1.26 | 600 | 0.6591 | -0.1979 | -0.2968 | 0.6640 | 0.0989 | -261.1124 | -280.0337 | 0.7883 | 0.6828 |
|
| 72 |
+
| 0.6488 | 1.47 | 700 | 0.6559 | -0.2310 | -0.3414 | 0.6620 | 0.1104 | -265.5783 | -283.3440 | 0.7549 | 0.6511 |
|
| 73 |
+
| 0.6449 | 1.67 | 800 | 0.6542 | -0.2518 | -0.3695 | 0.6560 | 0.1177 | -268.3814 | -285.4226 | 0.7372 | 0.6347 |
|
| 74 |
+
| 0.6487 | 1.88 | 900 | 0.6539 | -0.2571 | -0.3764 | 0.6560 | 0.1193 | -269.0724 | -285.9532 | 0.7320 | 0.6299 |
|
| 75 |
+
|
| 76 |
+
|
| 77 |
+
### Framework versions
|
| 78 |
+
|
| 79 |
+
- PEFT 0.7.1
|
| 80 |
+
- Transformers 4.36.2
|
| 81 |
+
- Pytorch 2.1.2+cu118
|
| 82 |
+
- Datasets 2.14.6
|
| 83 |
+
- Tokenizers 0.15.2
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 41977616
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:18e8c7debd8dff51d1c2995b62bb81a4cff15719362ed2577f580e107afb4a6a
|
| 3 |
size 41977616
|
all_results.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"eval_logits/chosen": 0.6308508515357971,
|
| 4 |
+
"eval_logits/rejected": 0.7334519028663635,
|
| 5 |
+
"eval_logps/chosen": -285.9486999511719,
|
| 6 |
+
"eval_logps/rejected": -269.1014404296875,
|
| 7 |
+
"eval_loss": 0.6537346243858337,
|
| 8 |
+
"eval_rewards/accuracies": 0.6579999923706055,
|
| 9 |
+
"eval_rewards/chosen": -0.2570453882217407,
|
| 10 |
+
"eval_rewards/margins": 0.11960798501968384,
|
| 11 |
+
"eval_rewards/rejected": -0.37665337324142456,
|
| 12 |
+
"eval_runtime": 325.7825,
|
| 13 |
+
"eval_samples": 2000,
|
| 14 |
+
"eval_samples_per_second": 6.139,
|
| 15 |
+
"eval_steps_per_second": 0.384,
|
| 16 |
+
"train_loss": 0.6680307045922589,
|
| 17 |
+
"train_runtime": 18174.3674,
|
| 18 |
+
"train_samples": 30567,
|
| 19 |
+
"train_samples_per_second": 3.364,
|
| 20 |
+
"train_steps_per_second": 0.052
|
| 21 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"eval_logits/chosen": 0.6308508515357971,
|
| 4 |
+
"eval_logits/rejected": 0.7334519028663635,
|
| 5 |
+
"eval_logps/chosen": -285.9486999511719,
|
| 6 |
+
"eval_logps/rejected": -269.1014404296875,
|
| 7 |
+
"eval_loss": 0.6537346243858337,
|
| 8 |
+
"eval_rewards/accuracies": 0.6579999923706055,
|
| 9 |
+
"eval_rewards/chosen": -0.2570453882217407,
|
| 10 |
+
"eval_rewards/margins": 0.11960798501968384,
|
| 11 |
+
"eval_rewards/rejected": -0.37665337324142456,
|
| 12 |
+
"eval_runtime": 325.7825,
|
| 13 |
+
"eval_samples": 2000,
|
| 14 |
+
"eval_samples_per_second": 6.139,
|
| 15 |
+
"eval_steps_per_second": 0.384
|
| 16 |
+
}
|
runs/Mar04_11-43-39_gpu4-119-4/events.out.tfevents.1709513183.gpu4-119-4.2293727.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9f31e217d6d987f66ffee394f6fd2e669ed3c0b75e199acfdfd9eafcf6276480
|
| 3 |
+
size 72568
|
runs/Mar04_11-43-39_gpu4-119-4/events.out.tfevents.1709531683.gpu4-119-4.2293727.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9c7ce6d34e46b08c034c6db625bfab99f376909a1884b93d2f5a80033bbd3e95
|
| 3 |
+
size 828
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"train_loss": 0.6680307045922589,
|
| 4 |
+
"train_runtime": 18174.3674,
|
| 5 |
+
"train_samples": 30567,
|
| 6 |
+
"train_samples_per_second": 3.364,
|
| 7 |
+
"train_steps_per_second": 0.052
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1518 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.9968602825745683,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 954,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 5.208333333333333e-08,
|
| 14 |
+
"logits/chosen": 0.952304482460022,
|
| 15 |
+
"logits/rejected": 0.5888463854789734,
|
| 16 |
+
"logps/chosen": -223.79486083984375,
|
| 17 |
+
"logps/rejected": -209.482666015625,
|
| 18 |
+
"loss": 0.6931,
|
| 19 |
+
"rewards/accuracies": 0.0,
|
| 20 |
+
"rewards/chosen": 0.0,
|
| 21 |
+
"rewards/margins": 0.0,
|
| 22 |
+
"rewards/rejected": 0.0,
|
| 23 |
+
"step": 1
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.02,
|
| 27 |
+
"learning_rate": 5.208333333333334e-07,
|
| 28 |
+
"logits/chosen": 0.836148738861084,
|
| 29 |
+
"logits/rejected": 0.8545415997505188,
|
| 30 |
+
"logps/chosen": -236.217529296875,
|
| 31 |
+
"logps/rejected": -221.92974853515625,
|
| 32 |
+
"loss": 0.6932,
|
| 33 |
+
"rewards/accuracies": 0.4444444477558136,
|
| 34 |
+
"rewards/chosen": 0.00045002170372754335,
|
| 35 |
+
"rewards/margins": 0.0005102179129607975,
|
| 36 |
+
"rewards/rejected": -6.019628563080914e-05,
|
| 37 |
+
"step": 10
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.04,
|
| 41 |
+
"learning_rate": 1.0416666666666667e-06,
|
| 42 |
+
"logits/chosen": 0.8339964151382446,
|
| 43 |
+
"logits/rejected": 0.9286314249038696,
|
| 44 |
+
"logps/chosen": -254.7573699951172,
|
| 45 |
+
"logps/rejected": -247.9052734375,
|
| 46 |
+
"loss": 0.6931,
|
| 47 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 48 |
+
"rewards/chosen": -0.0004581348621286452,
|
| 49 |
+
"rewards/margins": 0.00032393785659223795,
|
| 50 |
+
"rewards/rejected": -0.0007820727187208831,
|
| 51 |
+
"step": 20
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.06,
|
| 55 |
+
"learning_rate": 1.5625e-06,
|
| 56 |
+
"logits/chosen": 0.8594380617141724,
|
| 57 |
+
"logits/rejected": 0.916830837726593,
|
| 58 |
+
"logps/chosen": -260.4742736816406,
|
| 59 |
+
"logps/rejected": -232.0736846923828,
|
| 60 |
+
"loss": 0.6929,
|
| 61 |
+
"rewards/accuracies": 0.53125,
|
| 62 |
+
"rewards/chosen": 0.0009176501189358532,
|
| 63 |
+
"rewards/margins": 0.0013712994987145066,
|
| 64 |
+
"rewards/rejected": -0.00045364940888248384,
|
| 65 |
+
"step": 30
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.08,
|
| 69 |
+
"learning_rate": 2.0833333333333334e-06,
|
| 70 |
+
"logits/chosen": 0.8113815188407898,
|
| 71 |
+
"logits/rejected": 0.9032772183418274,
|
| 72 |
+
"logps/chosen": -280.20867919921875,
|
| 73 |
+
"logps/rejected": -228.60598754882812,
|
| 74 |
+
"loss": 0.693,
|
| 75 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 76 |
+
"rewards/chosen": 3.258117794757709e-05,
|
| 77 |
+
"rewards/margins": 0.0006167444516904652,
|
| 78 |
+
"rewards/rejected": -0.0005841633537784219,
|
| 79 |
+
"step": 40
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.1,
|
| 83 |
+
"learning_rate": 2.604166666666667e-06,
|
| 84 |
+
"logits/chosen": 0.8574679493904114,
|
| 85 |
+
"logits/rejected": 0.9363411068916321,
|
| 86 |
+
"logps/chosen": -257.6195373535156,
|
| 87 |
+
"logps/rejected": -219.19448852539062,
|
| 88 |
+
"loss": 0.6928,
|
| 89 |
+
"rewards/accuracies": 0.59375,
|
| 90 |
+
"rewards/chosen": 0.0007853487622924149,
|
| 91 |
+
"rewards/margins": 0.001927538076415658,
|
| 92 |
+
"rewards/rejected": -0.0011421891395002604,
|
| 93 |
+
"step": 50
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.13,
|
| 97 |
+
"learning_rate": 3.125e-06,
|
| 98 |
+
"logits/chosen": 0.8745417594909668,
|
| 99 |
+
"logits/rejected": 0.901624858379364,
|
| 100 |
+
"logps/chosen": -237.000244140625,
|
| 101 |
+
"logps/rejected": -237.52395629882812,
|
| 102 |
+
"loss": 0.6929,
|
| 103 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 104 |
+
"rewards/chosen": 1.0722555089159869e-05,
|
| 105 |
+
"rewards/margins": 0.0009478582069277763,
|
| 106 |
+
"rewards/rejected": -0.0009371357737109065,
|
| 107 |
+
"step": 60
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.15,
|
| 111 |
+
"learning_rate": 3.6458333333333333e-06,
|
| 112 |
+
"logits/chosen": 0.8507300615310669,
|
| 113 |
+
"logits/rejected": 0.8796240091323853,
|
| 114 |
+
"logps/chosen": -260.81085205078125,
|
| 115 |
+
"logps/rejected": -227.607421875,
|
| 116 |
+
"loss": 0.6931,
|
| 117 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 118 |
+
"rewards/chosen": -0.0008873953483998775,
|
| 119 |
+
"rewards/margins": -0.0005922773270867765,
|
| 120 |
+
"rewards/rejected": -0.00029511802131310105,
|
| 121 |
+
"step": 70
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.17,
|
| 125 |
+
"learning_rate": 4.166666666666667e-06,
|
| 126 |
+
"logits/chosen": 0.8868101239204407,
|
| 127 |
+
"logits/rejected": 0.9193674921989441,
|
| 128 |
+
"logps/chosen": -251.10498046875,
|
| 129 |
+
"logps/rejected": -231.2660675048828,
|
| 130 |
+
"loss": 0.6929,
|
| 131 |
+
"rewards/accuracies": 0.518750011920929,
|
| 132 |
+
"rewards/chosen": -0.00020074064377695322,
|
| 133 |
+
"rewards/margins": 0.0003954143321607262,
|
| 134 |
+
"rewards/rejected": -0.0005961551214568317,
|
| 135 |
+
"step": 80
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.19,
|
| 139 |
+
"learning_rate": 4.6875000000000004e-06,
|
| 140 |
+
"logits/chosen": 0.8633348345756531,
|
| 141 |
+
"logits/rejected": 0.9175950884819031,
|
| 142 |
+
"logps/chosen": -225.3305206298828,
|
| 143 |
+
"logps/rejected": -241.3758087158203,
|
| 144 |
+
"loss": 0.6931,
|
| 145 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 146 |
+
"rewards/chosen": -0.0006292141624726355,
|
| 147 |
+
"rewards/margins": 0.0004391434194985777,
|
| 148 |
+
"rewards/rejected": -0.0010683576110750437,
|
| 149 |
+
"step": 90
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.21,
|
| 153 |
+
"learning_rate": 4.999731868769027e-06,
|
| 154 |
+
"logits/chosen": 0.9241905212402344,
|
| 155 |
+
"logits/rejected": 0.9234986305236816,
|
| 156 |
+
"logps/chosen": -242.17343139648438,
|
| 157 |
+
"logps/rejected": -221.4640350341797,
|
| 158 |
+
"loss": 0.6929,
|
| 159 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 160 |
+
"rewards/chosen": -0.00037607658305205405,
|
| 161 |
+
"rewards/margins": 0.0006910150987096131,
|
| 162 |
+
"rewards/rejected": -0.0010670917108654976,
|
| 163 |
+
"step": 100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.21,
|
| 167 |
+
"eval_logits/chosen": 0.8145024180412292,
|
| 168 |
+
"eval_logits/rejected": 0.9167963862419128,
|
| 169 |
+
"eval_logps/chosen": -260.2240295410156,
|
| 170 |
+
"eval_logps/rejected": -231.53602600097656,
|
| 171 |
+
"eval_loss": 0.6927996873855591,
|
| 172 |
+
"eval_rewards/accuracies": 0.5320000052452087,
|
| 173 |
+
"eval_rewards/chosen": 0.00020143474102951586,
|
| 174 |
+
"eval_rewards/margins": 0.0012005382450297475,
|
| 175 |
+
"eval_rewards/rejected": -0.000999103649519384,
|
| 176 |
+
"eval_runtime": 326.2051,
|
| 177 |
+
"eval_samples_per_second": 6.131,
|
| 178 |
+
"eval_steps_per_second": 0.383,
|
| 179 |
+
"step": 100
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.23,
|
| 183 |
+
"learning_rate": 4.996716052911017e-06,
|
| 184 |
+
"logits/chosen": 0.8340839147567749,
|
| 185 |
+
"logits/rejected": 0.8756756782531738,
|
| 186 |
+
"logps/chosen": -264.01495361328125,
|
| 187 |
+
"logps/rejected": -219.5539093017578,
|
| 188 |
+
"loss": 0.6927,
|
| 189 |
+
"rewards/accuracies": 0.543749988079071,
|
| 190 |
+
"rewards/chosen": 7.205537986010313e-05,
|
| 191 |
+
"rewards/margins": 0.0008332778816111386,
|
| 192 |
+
"rewards/rejected": -0.0007612224435433745,
|
| 193 |
+
"step": 110
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.25,
|
| 197 |
+
"learning_rate": 4.9903533134293035e-06,
|
| 198 |
+
"logits/chosen": 0.8605300784111023,
|
| 199 |
+
"logits/rejected": 0.9697392582893372,
|
| 200 |
+
"logps/chosen": -254.9059295654297,
|
| 201 |
+
"logps/rejected": -219.1450958251953,
|
| 202 |
+
"loss": 0.6924,
|
| 203 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 204 |
+
"rewards/chosen": 0.0002610751544125378,
|
| 205 |
+
"rewards/margins": 0.002276130486279726,
|
| 206 |
+
"rewards/rejected": -0.002015055390074849,
|
| 207 |
+
"step": 120
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.27,
|
| 211 |
+
"learning_rate": 4.9806521797692184e-06,
|
| 212 |
+
"logits/chosen": 0.877278208732605,
|
| 213 |
+
"logits/rejected": 0.8787814974784851,
|
| 214 |
+
"logps/chosen": -264.6486511230469,
|
| 215 |
+
"logps/rejected": -247.06497192382812,
|
| 216 |
+
"loss": 0.6921,
|
| 217 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 218 |
+
"rewards/chosen": -0.00013167767610866576,
|
| 219 |
+
"rewards/margins": 0.002138237003237009,
|
| 220 |
+
"rewards/rejected": -0.0022699148394167423,
|
| 221 |
+
"step": 130
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.29,
|
| 225 |
+
"learning_rate": 4.967625656594782e-06,
|
| 226 |
+
"logits/chosen": 0.8538626432418823,
|
| 227 |
+
"logits/rejected": 0.9312636256217957,
|
| 228 |
+
"logps/chosen": -222.04147338867188,
|
| 229 |
+
"logps/rejected": -232.23892211914062,
|
| 230 |
+
"loss": 0.6921,
|
| 231 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 232 |
+
"rewards/chosen": -0.0011232274118810892,
|
| 233 |
+
"rewards/margins": 0.0009854320669546723,
|
| 234 |
+
"rewards/rejected": -0.0021086593624204397,
|
| 235 |
+
"step": 140
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.31,
|
| 239 |
+
"learning_rate": 4.95129120635556e-06,
|
| 240 |
+
"logits/chosen": 0.8747261166572571,
|
| 241 |
+
"logits/rejected": 0.9058718681335449,
|
| 242 |
+
"logps/chosen": -258.2791442871094,
|
| 243 |
+
"logps/rejected": -215.7833251953125,
|
| 244 |
+
"loss": 0.6919,
|
| 245 |
+
"rewards/accuracies": 0.5625,
|
| 246 |
+
"rewards/chosen": -0.0004979403456673026,
|
| 247 |
+
"rewards/margins": 0.002242505783215165,
|
| 248 |
+
"rewards/rejected": -0.0027404462452977896,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.33,
|
| 253 |
+
"learning_rate": 4.93167072587771e-06,
|
| 254 |
+
"logits/chosen": 0.7788752317428589,
|
| 255 |
+
"logits/rejected": 0.8470233082771301,
|
| 256 |
+
"logps/chosen": -257.824951171875,
|
| 257 |
+
"logps/rejected": -250.79281616210938,
|
| 258 |
+
"loss": 0.6914,
|
| 259 |
+
"rewards/accuracies": 0.606249988079071,
|
| 260 |
+
"rewards/chosen": -0.0017024253029376268,
|
| 261 |
+
"rewards/margins": 0.0027075002435594797,
|
| 262 |
+
"rewards/rejected": -0.0044099255464971066,
|
| 263 |
+
"step": 160
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.36,
|
| 267 |
+
"learning_rate": 4.908790517010637e-06,
|
| 268 |
+
"logits/chosen": 0.9306305050849915,
|
| 269 |
+
"logits/rejected": 0.9719738960266113,
|
| 270 |
+
"logps/chosen": -239.07101440429688,
|
| 271 |
+
"logps/rejected": -252.9778289794922,
|
| 272 |
+
"loss": 0.6911,
|
| 273 |
+
"rewards/accuracies": 0.550000011920929,
|
| 274 |
+
"rewards/chosen": -0.002183904405683279,
|
| 275 |
+
"rewards/margins": 0.002954904455691576,
|
| 276 |
+
"rewards/rejected": -0.005138809327036142,
|
| 277 |
+
"step": 170
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.38,
|
| 281 |
+
"learning_rate": 4.882681251368549e-06,
|
| 282 |
+
"logits/chosen": 0.853600800037384,
|
| 283 |
+
"logits/rejected": 0.8488330841064453,
|
| 284 |
+
"logps/chosen": -270.89105224609375,
|
| 285 |
+
"logps/rejected": -257.0163269042969,
|
| 286 |
+
"loss": 0.6906,
|
| 287 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 288 |
+
"rewards/chosen": -0.0012318979715928435,
|
| 289 |
+
"rewards/margins": 0.004222923889756203,
|
| 290 |
+
"rewards/rejected": -0.0054548210464417934,
|
| 291 |
+
"step": 180
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.4,
|
| 295 |
+
"learning_rate": 4.853377929214243e-06,
|
| 296 |
+
"logits/chosen": 0.7871135473251343,
|
| 297 |
+
"logits/rejected": 0.8603521585464478,
|
| 298 |
+
"logps/chosen": -252.1376190185547,
|
| 299 |
+
"logps/rejected": -245.2152862548828,
|
| 300 |
+
"loss": 0.6898,
|
| 301 |
+
"rewards/accuracies": 0.625,
|
| 302 |
+
"rewards/chosen": -0.0007233443320728838,
|
| 303 |
+
"rewards/margins": 0.007929561659693718,
|
| 304 |
+
"rewards/rejected": -0.008652905933558941,
|
| 305 |
+
"step": 190
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.42,
|
| 309 |
+
"learning_rate": 4.8209198325401815e-06,
|
| 310 |
+
"logits/chosen": 0.8782480955123901,
|
| 311 |
+
"logits/rejected": 0.8951404690742493,
|
| 312 |
+
"logps/chosen": -237.61087036132812,
|
| 313 |
+
"logps/rejected": -233.7824249267578,
|
| 314 |
+
"loss": 0.6893,
|
| 315 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 316 |
+
"rewards/chosen": -0.002221657894551754,
|
| 317 |
+
"rewards/margins": 0.00844704918563366,
|
| 318 |
+
"rewards/rejected": -0.010668707080185413,
|
| 319 |
+
"step": 200
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.42,
|
| 323 |
+
"eval_logits/chosen": 0.8204554915428162,
|
| 324 |
+
"eval_logits/rejected": 0.9234155416488647,
|
| 325 |
+
"eval_logps/chosen": -260.62249755859375,
|
| 326 |
+
"eval_logps/rejected": -232.77418518066406,
|
| 327 |
+
"eval_loss": 0.6890601515769958,
|
| 328 |
+
"eval_rewards/accuracies": 0.6499999761581421,
|
| 329 |
+
"eval_rewards/chosen": -0.0037833875976502895,
|
| 330 |
+
"eval_rewards/margins": 0.009597329422831535,
|
| 331 |
+
"eval_rewards/rejected": -0.013380718417465687,
|
| 332 |
+
"eval_runtime": 326.0716,
|
| 333 |
+
"eval_samples_per_second": 6.134,
|
| 334 |
+
"eval_steps_per_second": 0.383,
|
| 335 |
+
"step": 200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 0.44,
|
| 339 |
+
"learning_rate": 4.785350472409792e-06,
|
| 340 |
+
"logits/chosen": 0.906767725944519,
|
| 341 |
+
"logits/rejected": 0.9123601913452148,
|
| 342 |
+
"logps/chosen": -233.3804168701172,
|
| 343 |
+
"logps/rejected": -228.94284057617188,
|
| 344 |
+
"loss": 0.689,
|
| 345 |
+
"rewards/accuracies": 0.606249988079071,
|
| 346 |
+
"rewards/chosen": -0.00779892411082983,
|
| 347 |
+
"rewards/margins": 0.007471003569662571,
|
| 348 |
+
"rewards/rejected": -0.015269925817847252,
|
| 349 |
+
"step": 210
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 0.46,
|
| 353 |
+
"learning_rate": 4.746717530629565e-06,
|
| 354 |
+
"logits/chosen": 0.8570274114608765,
|
| 355 |
+
"logits/rejected": 0.8925272226333618,
|
| 356 |
+
"logps/chosen": -259.993896484375,
|
| 357 |
+
"logps/rejected": -239.8516082763672,
|
| 358 |
+
"loss": 0.6887,
|
| 359 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 360 |
+
"rewards/chosen": -0.006393183022737503,
|
| 361 |
+
"rewards/margins": 0.01187940128147602,
|
| 362 |
+
"rewards/rejected": -0.018272582441568375,
|
| 363 |
+
"step": 220
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.48,
|
| 367 |
+
"learning_rate": 4.7050727958301505e-06,
|
| 368 |
+
"logits/chosen": 0.8952150344848633,
|
| 369 |
+
"logits/rejected": 0.8783475756645203,
|
| 370 |
+
"logps/chosen": -244.82919311523438,
|
| 371 |
+
"logps/rejected": -230.5172882080078,
|
| 372 |
+
"loss": 0.687,
|
| 373 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 374 |
+
"rewards/chosen": -0.008937228471040726,
|
| 375 |
+
"rewards/margins": 0.014644329436123371,
|
| 376 |
+
"rewards/rejected": -0.023581556975841522,
|
| 377 |
+
"step": 230
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.5,
|
| 381 |
+
"learning_rate": 4.660472094042121e-06,
|
| 382 |
+
"logits/chosen": 0.857439398765564,
|
| 383 |
+
"logits/rejected": 0.8656377792358398,
|
| 384 |
+
"logps/chosen": -282.5680847167969,
|
| 385 |
+
"logps/rejected": -236.55111694335938,
|
| 386 |
+
"loss": 0.6853,
|
| 387 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 388 |
+
"rewards/chosen": -0.012710051611065865,
|
| 389 |
+
"rewards/margins": 0.017861289903521538,
|
| 390 |
+
"rewards/rejected": -0.030571341514587402,
|
| 391 |
+
"step": 240
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.52,
|
| 395 |
+
"learning_rate": 4.612975213859487e-06,
|
| 396 |
+
"logits/chosen": 0.8455727696418762,
|
| 397 |
+
"logits/rejected": 0.9185819625854492,
|
| 398 |
+
"logps/chosen": -268.9493713378906,
|
| 399 |
+
"logps/rejected": -243.8383331298828,
|
| 400 |
+
"loss": 0.6852,
|
| 401 |
+
"rewards/accuracies": 0.643750011920929,
|
| 402 |
+
"rewards/chosen": -0.014121539890766144,
|
| 403 |
+
"rewards/margins": 0.015637289732694626,
|
| 404 |
+
"rewards/rejected": -0.02975882962346077,
|
| 405 |
+
"step": 250
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.54,
|
| 409 |
+
"learning_rate": 4.5626458262912745e-06,
|
| 410 |
+
"logits/chosen": 0.858254075050354,
|
| 411 |
+
"logits/rejected": 0.8988651037216187,
|
| 412 |
+
"logps/chosen": -274.9979553222656,
|
| 413 |
+
"logps/rejected": -262.4846496582031,
|
| 414 |
+
"loss": 0.6842,
|
| 415 |
+
"rewards/accuracies": 0.675000011920929,
|
| 416 |
+
"rewards/chosen": -0.008313321508467197,
|
| 417 |
+
"rewards/margins": 0.024480264633893967,
|
| 418 |
+
"rewards/rejected": -0.03279358521103859,
|
| 419 |
+
"step": 260
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 0.57,
|
| 423 |
+
"learning_rate": 4.509551399408598e-06,
|
| 424 |
+
"logits/chosen": 0.9360445737838745,
|
| 425 |
+
"logits/rejected": 0.9437822103500366,
|
| 426 |
+
"logps/chosen": -253.78689575195312,
|
| 427 |
+
"logps/rejected": -211.69509887695312,
|
| 428 |
+
"loss": 0.683,
|
| 429 |
+
"rewards/accuracies": 0.699999988079071,
|
| 430 |
+
"rewards/chosen": -0.015140311792492867,
|
| 431 |
+
"rewards/margins": 0.022892246022820473,
|
| 432 |
+
"rewards/rejected": -0.03803255409002304,
|
| 433 |
+
"step": 270
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.59,
|
| 437 |
+
"learning_rate": 4.453763107901676e-06,
|
| 438 |
+
"logits/chosen": 0.9350343942642212,
|
| 439 |
+
"logits/rejected": 0.8994883298873901,
|
| 440 |
+
"logps/chosen": -247.9213104248047,
|
| 441 |
+
"logps/rejected": -256.08056640625,
|
| 442 |
+
"loss": 0.6842,
|
| 443 |
+
"rewards/accuracies": 0.612500011920929,
|
| 444 |
+
"rewards/chosen": -0.025496864691376686,
|
| 445 |
+
"rewards/margins": 0.012935856357216835,
|
| 446 |
+
"rewards/rejected": -0.03843272104859352,
|
| 447 |
+
"step": 280
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.61,
|
| 451 |
+
"learning_rate": 4.3953557376679856e-06,
|
| 452 |
+
"logits/chosen": 0.8583111763000488,
|
| 453 |
+
"logits/rejected": 0.8643442988395691,
|
| 454 |
+
"logps/chosen": -262.2920837402344,
|
| 455 |
+
"logps/rejected": -258.9486083984375,
|
| 456 |
+
"loss": 0.6827,
|
| 457 |
+
"rewards/accuracies": 0.59375,
|
| 458 |
+
"rewards/chosen": -0.027247220277786255,
|
| 459 |
+
"rewards/margins": 0.015835126861929893,
|
| 460 |
+
"rewards/rejected": -0.0430823490023613,
|
| 461 |
+
"step": 290
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.63,
|
| 465 |
+
"learning_rate": 4.33440758555951e-06,
|
| 466 |
+
"logits/chosen": 0.8355463743209839,
|
| 467 |
+
"logits/rejected": 0.9161020517349243,
|
| 468 |
+
"logps/chosen": -250.9317626953125,
|
| 469 |
+
"logps/rejected": -247.6198272705078,
|
| 470 |
+
"loss": 0.6809,
|
| 471 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 472 |
+
"rewards/chosen": -0.018615344539284706,
|
| 473 |
+
"rewards/margins": 0.033209316432476044,
|
| 474 |
+
"rewards/rejected": -0.0518246591091156,
|
| 475 |
+
"step": 300
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.63,
|
| 479 |
+
"eval_logits/chosen": 0.8092363476753235,
|
| 480 |
+
"eval_logits/rejected": 0.9150914549827576,
|
| 481 |
+
"eval_logps/chosen": -263.3646545410156,
|
| 482 |
+
"eval_logps/rejected": -237.54312133789062,
|
| 483 |
+
"eval_loss": 0.6809768080711365,
|
| 484 |
+
"eval_rewards/accuracies": 0.6679999828338623,
|
| 485 |
+
"eval_rewards/chosen": -0.03120502457022667,
|
| 486 |
+
"eval_rewards/margins": 0.029865048825740814,
|
| 487 |
+
"eval_rewards/rejected": -0.061070073395967484,
|
| 488 |
+
"eval_runtime": 325.9788,
|
| 489 |
+
"eval_samples_per_second": 6.135,
|
| 490 |
+
"eval_steps_per_second": 0.383,
|
| 491 |
+
"step": 300
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.65,
|
| 495 |
+
"learning_rate": 4.2710003544234255e-06,
|
| 496 |
+
"logits/chosen": 0.855351448059082,
|
| 497 |
+
"logits/rejected": 0.8703336715698242,
|
| 498 |
+
"logps/chosen": -238.54061889648438,
|
| 499 |
+
"logps/rejected": -230.7489776611328,
|
| 500 |
+
"loss": 0.6787,
|
| 501 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 502 |
+
"rewards/chosen": -0.036315977573394775,
|
| 503 |
+
"rewards/margins": 0.025386247783899307,
|
| 504 |
+
"rewards/rejected": -0.061702221632003784,
|
| 505 |
+
"step": 310
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.67,
|
| 509 |
+
"learning_rate": 4.205219043576955e-06,
|
| 510 |
+
"logits/chosen": 0.8615506887435913,
|
| 511 |
+
"logits/rejected": 0.8854458928108215,
|
| 512 |
+
"logps/chosen": -227.1722869873047,
|
| 513 |
+
"logps/rejected": -220.97412109375,
|
| 514 |
+
"loss": 0.6785,
|
| 515 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 516 |
+
"rewards/chosen": -0.03975047916173935,
|
| 517 |
+
"rewards/margins": 0.028606727719306946,
|
| 518 |
+
"rewards/rejected": -0.06835721433162689,
|
| 519 |
+
"step": 320
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.69,
|
| 523 |
+
"learning_rate": 4.137151834863213e-06,
|
| 524 |
+
"logits/chosen": 0.8440972566604614,
|
| 525 |
+
"logits/rejected": 0.8271803855895996,
|
| 526 |
+
"logps/chosen": -256.4277038574219,
|
| 527 |
+
"logps/rejected": -264.7651062011719,
|
| 528 |
+
"loss": 0.6787,
|
| 529 |
+
"rewards/accuracies": 0.637499988079071,
|
| 530 |
+
"rewards/chosen": -0.04135146737098694,
|
| 531 |
+
"rewards/margins": 0.039213813841342926,
|
| 532 |
+
"rewards/rejected": -0.08056528866291046,
|
| 533 |
+
"step": 330
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.71,
|
| 537 |
+
"learning_rate": 4.066889974440757e-06,
|
| 538 |
+
"logits/chosen": 0.8700094223022461,
|
| 539 |
+
"logits/rejected": 0.8766329884529114,
|
| 540 |
+
"logps/chosen": -269.4642333984375,
|
| 541 |
+
"logps/rejected": -255.2469024658203,
|
| 542 |
+
"loss": 0.6737,
|
| 543 |
+
"rewards/accuracies": 0.65625,
|
| 544 |
+
"rewards/chosen": -0.03877495974302292,
|
| 545 |
+
"rewards/margins": 0.04113053157925606,
|
| 546 |
+
"rewards/rejected": -0.07990548759698868,
|
| 547 |
+
"step": 340
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 0.73,
|
| 551 |
+
"learning_rate": 3.994527650465352e-06,
|
| 552 |
+
"logits/chosen": 0.8648789525032043,
|
| 553 |
+
"logits/rejected": 0.8878329396247864,
|
| 554 |
+
"logps/chosen": -227.1282958984375,
|
| 555 |
+
"logps/rejected": -212.8804931640625,
|
| 556 |
+
"loss": 0.6755,
|
| 557 |
+
"rewards/accuracies": 0.6875,
|
| 558 |
+
"rewards/chosen": -0.05504993349313736,
|
| 559 |
+
"rewards/margins": 0.040687672793865204,
|
| 560 |
+
"rewards/rejected": -0.09573759883642197,
|
| 561 |
+
"step": 350
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.75,
|
| 565 |
+
"learning_rate": 3.92016186682789e-06,
|
| 566 |
+
"logits/chosen": 0.7903240919113159,
|
| 567 |
+
"logits/rejected": 0.8026289939880371,
|
| 568 |
+
"logps/chosen": -214.0880889892578,
|
| 569 |
+
"logps/rejected": -244.24618530273438,
|
| 570 |
+
"loss": 0.6752,
|
| 571 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 572 |
+
"rewards/chosen": -0.06658105552196503,
|
| 573 |
+
"rewards/margins": 0.03603743016719818,
|
| 574 |
+
"rewards/rejected": -0.1026184931397438,
|
| 575 |
+
"step": 360
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.77,
|
| 579 |
+
"learning_rate": 3.843892313117724e-06,
|
| 580 |
+
"logits/chosen": 0.8764356374740601,
|
| 581 |
+
"logits/rejected": 0.9107359051704407,
|
| 582 |
+
"logps/chosen": -273.88873291015625,
|
| 583 |
+
"logps/rejected": -251.10812377929688,
|
| 584 |
+
"loss": 0.6758,
|
| 585 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 586 |
+
"rewards/chosen": -0.06244384124875069,
|
| 587 |
+
"rewards/margins": 0.04542430490255356,
|
| 588 |
+
"rewards/rejected": -0.10786814987659454,
|
| 589 |
+
"step": 370
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 0.8,
|
| 593 |
+
"learning_rate": 3.7658212309857576e-06,
|
| 594 |
+
"logits/chosen": 0.8514490127563477,
|
| 595 |
+
"logits/rejected": 0.8745658993721008,
|
| 596 |
+
"logps/chosen": -244.4405059814453,
|
| 597 |
+
"logps/rejected": -224.3061981201172,
|
| 598 |
+
"loss": 0.6708,
|
| 599 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 600 |
+
"rewards/chosen": -0.06743437796831131,
|
| 601 |
+
"rewards/margins": 0.048164453357458115,
|
| 602 |
+
"rewards/rejected": -0.11559884250164032,
|
| 603 |
+
"step": 380
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.82,
|
| 607 |
+
"learning_rate": 3.686053277086401e-06,
|
| 608 |
+
"logits/chosen": 0.8000567555427551,
|
| 609 |
+
"logits/rejected": 0.8884351849555969,
|
| 610 |
+
"logps/chosen": -268.4157409667969,
|
| 611 |
+
"logps/rejected": -247.18115234375,
|
| 612 |
+
"loss": 0.6696,
|
| 613 |
+
"rewards/accuracies": 0.675000011920929,
|
| 614 |
+
"rewards/chosen": -0.07314668595790863,
|
| 615 |
+
"rewards/margins": 0.049585141241550446,
|
| 616 |
+
"rewards/rejected": -0.12273182719945908,
|
| 617 |
+
"step": 390
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.84,
|
| 621 |
+
"learning_rate": 3.604695382782159e-06,
|
| 622 |
+
"logits/chosen": 0.8107595443725586,
|
| 623 |
+
"logits/rejected": 0.8327080607414246,
|
| 624 |
+
"logps/chosen": -284.17218017578125,
|
| 625 |
+
"logps/rejected": -263.6735534667969,
|
| 626 |
+
"loss": 0.6671,
|
| 627 |
+
"rewards/accuracies": 0.643750011920929,
|
| 628 |
+
"rewards/chosen": -0.0770660936832428,
|
| 629 |
+
"rewards/margins": 0.0486028790473938,
|
| 630 |
+
"rewards/rejected": -0.1256689727306366,
|
| 631 |
+
"step": 400
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 0.84,
|
| 635 |
+
"eval_logits/chosen": 0.7712985873222351,
|
| 636 |
+
"eval_logits/rejected": 0.8790242671966553,
|
| 637 |
+
"eval_logps/chosen": -268.7867431640625,
|
| 638 |
+
"eval_logps/rejected": -245.51242065429688,
|
| 639 |
+
"eval_loss": 0.6722846627235413,
|
| 640 |
+
"eval_rewards/accuracies": 0.6639999747276306,
|
| 641 |
+
"eval_rewards/chosen": -0.0854262262582779,
|
| 642 |
+
"eval_rewards/margins": 0.055336710065603256,
|
| 643 |
+
"eval_rewards/rejected": -0.14076292514801025,
|
| 644 |
+
"eval_runtime": 325.7205,
|
| 645 |
+
"eval_samples_per_second": 6.14,
|
| 646 |
+
"eval_steps_per_second": 0.384,
|
| 647 |
+
"step": 400
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.86,
|
| 651 |
+
"learning_rate": 3.5218566107988872e-06,
|
| 652 |
+
"logits/chosen": 0.8288625478744507,
|
| 653 |
+
"logits/rejected": 0.8551315069198608,
|
| 654 |
+
"logps/chosen": -276.74462890625,
|
| 655 |
+
"logps/rejected": -247.576416015625,
|
| 656 |
+
"loss": 0.6718,
|
| 657 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 658 |
+
"rewards/chosen": -0.08322989195585251,
|
| 659 |
+
"rewards/margins": 0.06066075712442398,
|
| 660 |
+
"rewards/rejected": -0.1438906490802765,
|
| 661 |
+
"step": 410
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.88,
|
| 665 |
+
"learning_rate": 3.437648009023905e-06,
|
| 666 |
+
"logits/chosen": 0.7773796916007996,
|
| 667 |
+
"logits/rejected": 0.8443312644958496,
|
| 668 |
+
"logps/chosen": -222.3345184326172,
|
| 669 |
+
"logps/rejected": -218.86825561523438,
|
| 670 |
+
"loss": 0.6696,
|
| 671 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 672 |
+
"rewards/chosen": -0.10068164020776749,
|
| 673 |
+
"rewards/margins": 0.04352349415421486,
|
| 674 |
+
"rewards/rejected": -0.14420512318611145,
|
| 675 |
+
"step": 420
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.9,
|
| 679 |
+
"learning_rate": 3.352182461642929e-06,
|
| 680 |
+
"logits/chosen": 0.8025181889533997,
|
| 681 |
+
"logits/rejected": 0.8646955490112305,
|
| 682 |
+
"logps/chosen": -243.1133270263672,
|
| 683 |
+
"logps/rejected": -232.02047729492188,
|
| 684 |
+
"loss": 0.6638,
|
| 685 |
+
"rewards/accuracies": 0.6875,
|
| 686 |
+
"rewards/chosen": -0.08811922371387482,
|
| 687 |
+
"rewards/margins": 0.06207479164004326,
|
| 688 |
+
"rewards/rejected": -0.15019401907920837,
|
| 689 |
+
"step": 430
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.92,
|
| 693 |
+
"learning_rate": 3.265574537815398e-06,
|
| 694 |
+
"logits/chosen": 0.8092749714851379,
|
| 695 |
+
"logits/rejected": 0.8513230085372925,
|
| 696 |
+
"logps/chosen": -288.71636962890625,
|
| 697 |
+
"logps/rejected": -256.467529296875,
|
| 698 |
+
"loss": 0.6676,
|
| 699 |
+
"rewards/accuracies": 0.6875,
|
| 700 |
+
"rewards/chosen": -0.0855972021818161,
|
| 701 |
+
"rewards/margins": 0.07402561604976654,
|
| 702 |
+
"rewards/rejected": -0.15962281823158264,
|
| 703 |
+
"step": 440
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.94,
|
| 707 |
+
"learning_rate": 3.177940338091043e-06,
|
| 708 |
+
"logits/chosen": 0.8230969309806824,
|
| 709 |
+
"logits/rejected": 0.8771616816520691,
|
| 710 |
+
"logps/chosen": -262.885009765625,
|
| 711 |
+
"logps/rejected": -236.41366577148438,
|
| 712 |
+
"loss": 0.6661,
|
| 713 |
+
"rewards/accuracies": 0.668749988079071,
|
| 714 |
+
"rewards/chosen": -0.10893376171588898,
|
| 715 |
+
"rewards/margins": 0.05591960996389389,
|
| 716 |
+
"rewards/rejected": -0.16485336422920227,
|
| 717 |
+
"step": 450
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.96,
|
| 721 |
+
"learning_rate": 3.089397338773569e-06,
|
| 722 |
+
"logits/chosen": 0.8003429174423218,
|
| 723 |
+
"logits/rejected": 0.8373273015022278,
|
| 724 |
+
"logps/chosen": -269.59130859375,
|
| 725 |
+
"logps/rejected": -243.3413543701172,
|
| 726 |
+
"loss": 0.6635,
|
| 727 |
+
"rewards/accuracies": 0.706250011920929,
|
| 728 |
+
"rewards/chosen": -0.12066192924976349,
|
| 729 |
+
"rewards/margins": 0.07056786119937897,
|
| 730 |
+
"rewards/rejected": -0.19122979044914246,
|
| 731 |
+
"step": 460
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.98,
|
| 735 |
+
"learning_rate": 3.0000642344401115e-06,
|
| 736 |
+
"logits/chosen": 0.7690973877906799,
|
| 737 |
+
"logits/rejected": 0.8160909414291382,
|
| 738 |
+
"logps/chosen": -250.65017700195312,
|
| 739 |
+
"logps/rejected": -241.54763793945312,
|
| 740 |
+
"loss": 0.6624,
|
| 741 |
+
"rewards/accuracies": 0.637499988079071,
|
| 742 |
+
"rewards/chosen": -0.12458320707082748,
|
| 743 |
+
"rewards/margins": 0.06558409333229065,
|
| 744 |
+
"rewards/rejected": -0.19016727805137634,
|
| 745 |
+
"step": 470
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 1.0,
|
| 749 |
+
"learning_rate": 2.9100607788275547e-06,
|
| 750 |
+
"logits/chosen": 0.8020121455192566,
|
| 751 |
+
"logits/rejected": 0.8617356419563293,
|
| 752 |
+
"logps/chosen": -256.21771240234375,
|
| 753 |
+
"logps/rejected": -253.4419708251953,
|
| 754 |
+
"loss": 0.6658,
|
| 755 |
+
"rewards/accuracies": 0.668749988079071,
|
| 756 |
+
"rewards/chosen": -0.1322631537914276,
|
| 757 |
+
"rewards/margins": 0.06032481789588928,
|
| 758 |
+
"rewards/rejected": -0.1925879716873169,
|
| 759 |
+
"step": 480
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 1.03,
|
| 763 |
+
"learning_rate": 2.8195076242990124e-06,
|
| 764 |
+
"logits/chosen": 0.7743502259254456,
|
| 765 |
+
"logits/rejected": 0.8271788358688354,
|
| 766 |
+
"logps/chosen": -252.33804321289062,
|
| 767 |
+
"logps/rejected": -235.423828125,
|
| 768 |
+
"loss": 0.6624,
|
| 769 |
+
"rewards/accuracies": 0.643750011920929,
|
| 770 |
+
"rewards/chosen": -0.14343151450157166,
|
| 771 |
+
"rewards/margins": 0.07069016247987747,
|
| 772 |
+
"rewards/rejected": -0.21412166953086853,
|
| 773 |
+
"step": 490
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.05,
|
| 777 |
+
"learning_rate": 2.72852616010567e-06,
|
| 778 |
+
"logits/chosen": 0.7551236748695374,
|
| 779 |
+
"logits/rejected": 0.7969120144844055,
|
| 780 |
+
"logps/chosen": -266.84375,
|
| 781 |
+
"logps/rejected": -254.37460327148438,
|
| 782 |
+
"loss": 0.6627,
|
| 783 |
+
"rewards/accuracies": 0.71875,
|
| 784 |
+
"rewards/chosen": -0.1373104304075241,
|
| 785 |
+
"rewards/margins": 0.07557874917984009,
|
| 786 |
+
"rewards/rejected": -0.212889164686203,
|
| 787 |
+
"step": 500
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.05,
|
| 791 |
+
"eval_logits/chosen": 0.7216536998748779,
|
| 792 |
+
"eval_logits/rejected": 0.8293780088424683,
|
| 793 |
+
"eval_logps/chosen": -275.1849365234375,
|
| 794 |
+
"eval_logps/rejected": -254.370361328125,
|
| 795 |
+
"eval_loss": 0.664509654045105,
|
| 796 |
+
"eval_rewards/accuracies": 0.6679999828338623,
|
| 797 |
+
"eval_rewards/chosen": -0.14940780401229858,
|
| 798 |
+
"eval_rewards/margins": 0.07993472367525101,
|
| 799 |
+
"eval_rewards/rejected": -0.2293425351381302,
|
| 800 |
+
"eval_runtime": 325.7214,
|
| 801 |
+
"eval_samples_per_second": 6.14,
|
| 802 |
+
"eval_steps_per_second": 0.384,
|
| 803 |
+
"step": 500
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 1.07,
|
| 807 |
+
"learning_rate": 2.637238349660819e-06,
|
| 808 |
+
"logits/chosen": 0.7647446393966675,
|
| 809 |
+
"logits/rejected": 0.8582034111022949,
|
| 810 |
+
"logps/chosen": -251.79904174804688,
|
| 811 |
+
"logps/rejected": -218.86587524414062,
|
| 812 |
+
"loss": 0.665,
|
| 813 |
+
"rewards/accuracies": 0.6875,
|
| 814 |
+
"rewards/chosen": -0.16131970286369324,
|
| 815 |
+
"rewards/margins": 0.06838177144527435,
|
| 816 |
+
"rewards/rejected": -0.2297014743089676,
|
| 817 |
+
"step": 510
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 1.09,
|
| 821 |
+
"learning_rate": 2.5457665670441937e-06,
|
| 822 |
+
"logits/chosen": 0.8116461038589478,
|
| 823 |
+
"logits/rejected": 0.8353972434997559,
|
| 824 |
+
"logps/chosen": -263.5601501464844,
|
| 825 |
+
"logps/rejected": -247.66909790039062,
|
| 826 |
+
"loss": 0.6591,
|
| 827 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 828 |
+
"rewards/chosen": -0.14197978377342224,
|
| 829 |
+
"rewards/margins": 0.09031648933887482,
|
| 830 |
+
"rewards/rejected": -0.23229627311229706,
|
| 831 |
+
"step": 520
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 1.11,
|
| 835 |
+
"learning_rate": 2.4542334329558075e-06,
|
| 836 |
+
"logits/chosen": 0.7308142781257629,
|
| 837 |
+
"logits/rejected": 0.7632499933242798,
|
| 838 |
+
"logps/chosen": -257.2229919433594,
|
| 839 |
+
"logps/rejected": -250.84469604492188,
|
| 840 |
+
"loss": 0.6607,
|
| 841 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 842 |
+
"rewards/chosen": -0.15741421282291412,
|
| 843 |
+
"rewards/margins": 0.07882841676473618,
|
| 844 |
+
"rewards/rejected": -0.2362426221370697,
|
| 845 |
+
"step": 530
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.13,
|
| 849 |
+
"learning_rate": 2.3627616503391813e-06,
|
| 850 |
+
"logits/chosen": 0.7235521674156189,
|
| 851 |
+
"logits/rejected": 0.7589094042778015,
|
| 852 |
+
"logps/chosen": -274.9698791503906,
|
| 853 |
+
"logps/rejected": -238.1727752685547,
|
| 854 |
+
"loss": 0.6614,
|
| 855 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 856 |
+
"rewards/chosen": -0.16767871379852295,
|
| 857 |
+
"rewards/margins": 0.07072968035936356,
|
| 858 |
+
"rewards/rejected": -0.2384084016084671,
|
| 859 |
+
"step": 540
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.15,
|
| 863 |
+
"learning_rate": 2.271473839894331e-06,
|
| 864 |
+
"logits/chosen": 0.7224727869033813,
|
| 865 |
+
"logits/rejected": 0.7499375939369202,
|
| 866 |
+
"logps/chosen": -283.6424560546875,
|
| 867 |
+
"logps/rejected": -272.5013122558594,
|
| 868 |
+
"loss": 0.6593,
|
| 869 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 870 |
+
"rewards/chosen": -0.17033258080482483,
|
| 871 |
+
"rewards/margins": 0.07838577777147293,
|
| 872 |
+
"rewards/rejected": -0.24871835112571716,
|
| 873 |
+
"step": 550
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 1.17,
|
| 877 |
+
"learning_rate": 2.1804923757009885e-06,
|
| 878 |
+
"logits/chosen": 0.6960411667823792,
|
| 879 |
+
"logits/rejected": 0.7274879813194275,
|
| 880 |
+
"logps/chosen": -269.6304016113281,
|
| 881 |
+
"logps/rejected": -250.3975830078125,
|
| 882 |
+
"loss": 0.6614,
|
| 883 |
+
"rewards/accuracies": 0.6875,
|
| 884 |
+
"rewards/chosen": -0.18492145836353302,
|
| 885 |
+
"rewards/margins": 0.07485240697860718,
|
| 886 |
+
"rewards/rejected": -0.259773850440979,
|
| 887 |
+
"step": 560
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 1.19,
|
| 891 |
+
"learning_rate": 2.089939221172446e-06,
|
| 892 |
+
"logits/chosen": 0.7146167755126953,
|
| 893 |
+
"logits/rejected": 0.7337481379508972,
|
| 894 |
+
"logps/chosen": -288.8470153808594,
|
| 895 |
+
"logps/rejected": -258.131591796875,
|
| 896 |
+
"loss": 0.664,
|
| 897 |
+
"rewards/accuracies": 0.675000011920929,
|
| 898 |
+
"rewards/chosen": -0.19552810490131378,
|
| 899 |
+
"rewards/margins": 0.09874384850263596,
|
| 900 |
+
"rewards/rejected": -0.29427194595336914,
|
| 901 |
+
"step": 570
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 1.21,
|
| 905 |
+
"learning_rate": 1.9999357655598894e-06,
|
| 906 |
+
"logits/chosen": 0.7304507493972778,
|
| 907 |
+
"logits/rejected": 0.7039176225662231,
|
| 908 |
+
"logps/chosen": -259.52008056640625,
|
| 909 |
+
"logps/rejected": -250.97146606445312,
|
| 910 |
+
"loss": 0.6591,
|
| 911 |
+
"rewards/accuracies": 0.668749988079071,
|
| 912 |
+
"rewards/chosen": -0.19664093852043152,
|
| 913 |
+
"rewards/margins": 0.058397077023983,
|
| 914 |
+
"rewards/rejected": -0.2550380527973175,
|
| 915 |
+
"step": 580
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 1.23,
|
| 919 |
+
"learning_rate": 1.9106026612264316e-06,
|
| 920 |
+
"logits/chosen": 0.7533366084098816,
|
| 921 |
+
"logits/rejected": 0.8193215131759644,
|
| 922 |
+
"logps/chosen": -236.71243286132812,
|
| 923 |
+
"logps/rejected": -238.93679809570312,
|
| 924 |
+
"loss": 0.653,
|
| 925 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 926 |
+
"rewards/chosen": -0.1914629340171814,
|
| 927 |
+
"rewards/margins": 0.07271396368741989,
|
| 928 |
+
"rewards/rejected": -0.2641769051551819,
|
| 929 |
+
"step": 590
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 1.26,
|
| 933 |
+
"learning_rate": 1.8220596619089576e-06,
|
| 934 |
+
"logits/chosen": 0.7307643890380859,
|
| 935 |
+
"logits/rejected": 0.7107076048851013,
|
| 936 |
+
"logps/chosen": -281.09710693359375,
|
| 937 |
+
"logps/rejected": -258.6228942871094,
|
| 938 |
+
"loss": 0.6476,
|
| 939 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 940 |
+
"rewards/chosen": -0.19486048817634583,
|
| 941 |
+
"rewards/margins": 0.10412123054265976,
|
| 942 |
+
"rewards/rejected": -0.2989817261695862,
|
| 943 |
+
"step": 600
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 1.26,
|
| 947 |
+
"eval_logits/chosen": 0.6828243136405945,
|
| 948 |
+
"eval_logits/rejected": 0.7883425354957581,
|
| 949 |
+
"eval_logps/chosen": -280.03369140625,
|
| 950 |
+
"eval_logps/rejected": -261.11236572265625,
|
| 951 |
+
"eval_loss": 0.659087061882019,
|
| 952 |
+
"eval_rewards/accuracies": 0.6639999747276306,
|
| 953 |
+
"eval_rewards/chosen": -0.1978950798511505,
|
| 954 |
+
"eval_rewards/margins": 0.09886746108531952,
|
| 955 |
+
"eval_rewards/rejected": -0.2967625558376312,
|
| 956 |
+
"eval_runtime": 325.9412,
|
| 957 |
+
"eval_samples_per_second": 6.136,
|
| 958 |
+
"eval_steps_per_second": 0.384,
|
| 959 |
+
"step": 600
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 1.28,
|
| 963 |
+
"learning_rate": 1.7344254621846018e-06,
|
| 964 |
+
"logits/chosen": 0.700169026851654,
|
| 965 |
+
"logits/rejected": 0.7478510141372681,
|
| 966 |
+
"logps/chosen": -283.26641845703125,
|
| 967 |
+
"logps/rejected": -278.6813049316406,
|
| 968 |
+
"loss": 0.6514,
|
| 969 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 970 |
+
"rewards/chosen": -0.191552072763443,
|
| 971 |
+
"rewards/margins": 0.09749529510736465,
|
| 972 |
+
"rewards/rejected": -0.28904733061790466,
|
| 973 |
+
"step": 610
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 1.3,
|
| 977 |
+
"learning_rate": 1.647817538357072e-06,
|
| 978 |
+
"logits/chosen": 0.7241575121879578,
|
| 979 |
+
"logits/rejected": 0.7943159341812134,
|
| 980 |
+
"logps/chosen": -267.4305725097656,
|
| 981 |
+
"logps/rejected": -256.3984069824219,
|
| 982 |
+
"loss": 0.6591,
|
| 983 |
+
"rewards/accuracies": 0.612500011920929,
|
| 984 |
+
"rewards/chosen": -0.2104833871126175,
|
| 985 |
+
"rewards/margins": 0.07614756375551224,
|
| 986 |
+
"rewards/rejected": -0.2866309583187103,
|
| 987 |
+
"step": 620
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 1.32,
|
| 991 |
+
"learning_rate": 1.5623519909760953e-06,
|
| 992 |
+
"logits/chosen": 0.6998089551925659,
|
| 993 |
+
"logits/rejected": 0.7587723731994629,
|
| 994 |
+
"logps/chosen": -264.82110595703125,
|
| 995 |
+
"logps/rejected": -267.1747741699219,
|
| 996 |
+
"loss": 0.6551,
|
| 997 |
+
"rewards/accuracies": 0.637499988079071,
|
| 998 |
+
"rewards/chosen": -0.22911302745342255,
|
| 999 |
+
"rewards/margins": 0.07119645923376083,
|
| 1000 |
+
"rewards/rejected": -0.30030950903892517,
|
| 1001 |
+
"step": 630
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 1.34,
|
| 1005 |
+
"learning_rate": 1.4781433892011132e-06,
|
| 1006 |
+
"logits/chosen": 0.7701424360275269,
|
| 1007 |
+
"logits/rejected": 0.7888699173927307,
|
| 1008 |
+
"logps/chosen": -262.2229309082031,
|
| 1009 |
+
"logps/rejected": -286.0958557128906,
|
| 1010 |
+
"loss": 0.658,
|
| 1011 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1012 |
+
"rewards/chosen": -0.2087526023387909,
|
| 1013 |
+
"rewards/margins": 0.07823307812213898,
|
| 1014 |
+
"rewards/rejected": -0.28698569536209106,
|
| 1015 |
+
"step": 640
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 1.36,
|
| 1019 |
+
"learning_rate": 1.3953046172178413e-06,
|
| 1020 |
+
"logits/chosen": 0.7079032063484192,
|
| 1021 |
+
"logits/rejected": 0.7274273037910461,
|
| 1022 |
+
"logps/chosen": -276.0498352050781,
|
| 1023 |
+
"logps/rejected": -277.2353820800781,
|
| 1024 |
+
"loss": 0.6536,
|
| 1025 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1026 |
+
"rewards/chosen": -0.23646607995033264,
|
| 1027 |
+
"rewards/margins": 0.09612749516963959,
|
| 1028 |
+
"rewards/rejected": -0.33259356021881104,
|
| 1029 |
+
"step": 650
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 1.38,
|
| 1033 |
+
"learning_rate": 1.3139467229135999e-06,
|
| 1034 |
+
"logits/chosen": 0.6826360821723938,
|
| 1035 |
+
"logits/rejected": 0.751067578792572,
|
| 1036 |
+
"logps/chosen": -241.80044555664062,
|
| 1037 |
+
"logps/rejected": -275.70989990234375,
|
| 1038 |
+
"loss": 0.6556,
|
| 1039 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1040 |
+
"rewards/chosen": -0.22094163298606873,
|
| 1041 |
+
"rewards/margins": 0.08276458084583282,
|
| 1042 |
+
"rewards/rejected": -0.30370622873306274,
|
| 1043 |
+
"step": 660
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 1.4,
|
| 1047 |
+
"learning_rate": 1.2341787690142436e-06,
|
| 1048 |
+
"logits/chosen": 0.6527940630912781,
|
| 1049 |
+
"logits/rejected": 0.7365330457687378,
|
| 1050 |
+
"logps/chosen": -320.76385498046875,
|
| 1051 |
+
"logps/rejected": -271.78790283203125,
|
| 1052 |
+
"loss": 0.6552,
|
| 1053 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1054 |
+
"rewards/chosen": -0.20969745516777039,
|
| 1055 |
+
"rewards/margins": 0.11965203285217285,
|
| 1056 |
+
"rewards/rejected": -0.32934948801994324,
|
| 1057 |
+
"step": 670
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 1.42,
|
| 1061 |
+
"learning_rate": 1.1561076868822756e-06,
|
| 1062 |
+
"logits/chosen": 0.6766701936721802,
|
| 1063 |
+
"logits/rejected": 0.7141178250312805,
|
| 1064 |
+
"logps/chosen": -265.7345886230469,
|
| 1065 |
+
"logps/rejected": -245.7202606201172,
|
| 1066 |
+
"loss": 0.6518,
|
| 1067 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1068 |
+
"rewards/chosen": -0.22580020129680634,
|
| 1069 |
+
"rewards/margins": 0.0855962485074997,
|
| 1070 |
+
"rewards/rejected": -0.31139642000198364,
|
| 1071 |
+
"step": 680
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 1.44,
|
| 1075 |
+
"learning_rate": 1.079838133172111e-06,
|
| 1076 |
+
"logits/chosen": 0.7425110936164856,
|
| 1077 |
+
"logits/rejected": 0.7384303212165833,
|
| 1078 |
+
"logps/chosen": -278.8888854980469,
|
| 1079 |
+
"logps/rejected": -266.8258972167969,
|
| 1080 |
+
"loss": 0.6598,
|
| 1081 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 1082 |
+
"rewards/chosen": -0.21307018399238586,
|
| 1083 |
+
"rewards/margins": 0.08619161695241928,
|
| 1084 |
+
"rewards/rejected": -0.29926180839538574,
|
| 1085 |
+
"step": 690
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 1.47,
|
| 1089 |
+
"learning_rate": 1.0054723495346484e-06,
|
| 1090 |
+
"logits/chosen": 0.6221153736114502,
|
| 1091 |
+
"logits/rejected": 0.6998614072799683,
|
| 1092 |
+
"logps/chosen": -280.8089599609375,
|
| 1093 |
+
"logps/rejected": -257.3545837402344,
|
| 1094 |
+
"loss": 0.6488,
|
| 1095 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1096 |
+
"rewards/chosen": -0.22230124473571777,
|
| 1097 |
+
"rewards/margins": 0.09477819502353668,
|
| 1098 |
+
"rewards/rejected": -0.31707945466041565,
|
| 1099 |
+
"step": 700
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 1.47,
|
| 1103 |
+
"eval_logits/chosen": 0.651095986366272,
|
| 1104 |
+
"eval_logits/rejected": 0.7549068927764893,
|
| 1105 |
+
"eval_logps/chosen": -283.3440246582031,
|
| 1106 |
+
"eval_logps/rejected": -265.57830810546875,
|
| 1107 |
+
"eval_loss": 0.6559058427810669,
|
| 1108 |
+
"eval_rewards/accuracies": 0.6620000004768372,
|
| 1109 |
+
"eval_rewards/chosen": -0.23099879920482635,
|
| 1110 |
+
"eval_rewards/margins": 0.11042327433824539,
|
| 1111 |
+
"eval_rewards/rejected": -0.34142205119132996,
|
| 1112 |
+
"eval_runtime": 325.7526,
|
| 1113 |
+
"eval_samples_per_second": 6.14,
|
| 1114 |
+
"eval_steps_per_second": 0.384,
|
| 1115 |
+
"step": 700
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 1.49,
|
| 1119 |
+
"learning_rate": 9.331100255592437e-07,
|
| 1120 |
+
"logits/chosen": 0.7064228653907776,
|
| 1121 |
+
"logits/rejected": 0.7716813683509827,
|
| 1122 |
+
"logps/chosen": -300.17694091796875,
|
| 1123 |
+
"logps/rejected": -261.4855651855469,
|
| 1124 |
+
"loss": 0.6475,
|
| 1125 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 1126 |
+
"rewards/chosen": -0.2144012451171875,
|
| 1127 |
+
"rewards/margins": 0.12820136547088623,
|
| 1128 |
+
"rewards/rejected": -0.34260261058807373,
|
| 1129 |
+
"step": 710
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 1.51,
|
| 1133 |
+
"learning_rate": 8.628481651367876e-07,
|
| 1134 |
+
"logits/chosen": 0.6913928985595703,
|
| 1135 |
+
"logits/rejected": 0.7301486730575562,
|
| 1136 |
+
"logps/chosen": -286.8462829589844,
|
| 1137 |
+
"logps/rejected": -276.6960754394531,
|
| 1138 |
+
"loss": 0.6457,
|
| 1139 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1140 |
+
"rewards/chosen": -0.23222024738788605,
|
| 1141 |
+
"rewards/margins": 0.10797496885061264,
|
| 1142 |
+
"rewards/rejected": -0.3401952087879181,
|
| 1143 |
+
"step": 720
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.53,
|
| 1147 |
+
"learning_rate": 7.947809564230446e-07,
|
| 1148 |
+
"logits/chosen": 0.7156798839569092,
|
| 1149 |
+
"logits/rejected": 0.7468422055244446,
|
| 1150 |
+
"logps/chosen": -288.5824890136719,
|
| 1151 |
+
"logps/rejected": -286.32293701171875,
|
| 1152 |
+
"loss": 0.6439,
|
| 1153 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1154 |
+
"rewards/chosen": -0.23393268883228302,
|
| 1155 |
+
"rewards/margins": 0.12296883016824722,
|
| 1156 |
+
"rewards/rejected": -0.35690149664878845,
|
| 1157 |
+
"step": 730
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 1.55,
|
| 1161 |
+
"learning_rate": 7.289996455765749e-07,
|
| 1162 |
+
"logits/chosen": 0.668700098991394,
|
| 1163 |
+
"logits/rejected": 0.6845619678497314,
|
| 1164 |
+
"logps/chosen": -299.6327819824219,
|
| 1165 |
+
"logps/rejected": -283.09295654296875,
|
| 1166 |
+
"loss": 0.6584,
|
| 1167 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1168 |
+
"rewards/chosen": -0.2437756061553955,
|
| 1169 |
+
"rewards/margins": 0.07863186299800873,
|
| 1170 |
+
"rewards/rejected": -0.32240745425224304,
|
| 1171 |
+
"step": 740
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 1.57,
|
| 1175 |
+
"learning_rate": 6.655924144404907e-07,
|
| 1176 |
+
"logits/chosen": 0.6234613656997681,
|
| 1177 |
+
"logits/rejected": 0.6867505311965942,
|
| 1178 |
+
"logps/chosen": -281.64007568359375,
|
| 1179 |
+
"logps/rejected": -268.5241394042969,
|
| 1180 |
+
"loss": 0.6515,
|
| 1181 |
+
"rewards/accuracies": 0.6875,
|
| 1182 |
+
"rewards/chosen": -0.21235065162181854,
|
| 1183 |
+
"rewards/margins": 0.12807723879814148,
|
| 1184 |
+
"rewards/rejected": -0.34042784571647644,
|
| 1185 |
+
"step": 750
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 1.59,
|
| 1189 |
+
"learning_rate": 6.046442623320145e-07,
|
| 1190 |
+
"logits/chosen": 0.7112501859664917,
|
| 1191 |
+
"logits/rejected": 0.7498377561569214,
|
| 1192 |
+
"logps/chosen": -281.40875244140625,
|
| 1193 |
+
"logps/rejected": -251.820068359375,
|
| 1194 |
+
"loss": 0.6496,
|
| 1195 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1196 |
+
"rewards/chosen": -0.26181793212890625,
|
| 1197 |
+
"rewards/margins": 0.08466657996177673,
|
| 1198 |
+
"rewards/rejected": -0.3464844822883606,
|
| 1199 |
+
"step": 760
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 1.61,
|
| 1203 |
+
"learning_rate": 5.462368920983249e-07,
|
| 1204 |
+
"logits/chosen": 0.6798941493034363,
|
| 1205 |
+
"logits/rejected": 0.7346147298812866,
|
| 1206 |
+
"logps/chosen": -266.92315673828125,
|
| 1207 |
+
"logps/rejected": -247.5948028564453,
|
| 1208 |
+
"loss": 0.6566,
|
| 1209 |
+
"rewards/accuracies": 0.625,
|
| 1210 |
+
"rewards/chosen": -0.25860121846199036,
|
| 1211 |
+
"rewards/margins": 0.05472899600863457,
|
| 1212 |
+
"rewards/rejected": -0.31333020329475403,
|
| 1213 |
+
"step": 770
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 1.63,
|
| 1217 |
+
"learning_rate": 4.904486005914027e-07,
|
| 1218 |
+
"logits/chosen": 0.6729373931884766,
|
| 1219 |
+
"logits/rejected": 0.7290818095207214,
|
| 1220 |
+
"logps/chosen": -307.30926513671875,
|
| 1221 |
+
"logps/rejected": -292.6742248535156,
|
| 1222 |
+
"loss": 0.6463,
|
| 1223 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1224 |
+
"rewards/chosen": -0.2319561243057251,
|
| 1225 |
+
"rewards/margins": 0.1268167346715927,
|
| 1226 |
+
"rewards/rejected": -0.3587728440761566,
|
| 1227 |
+
"step": 780
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 1.65,
|
| 1231 |
+
"learning_rate": 4.373541737087264e-07,
|
| 1232 |
+
"logits/chosen": 0.6597692966461182,
|
| 1233 |
+
"logits/rejected": 0.7631260752677917,
|
| 1234 |
+
"logps/chosen": -288.7844543457031,
|
| 1235 |
+
"logps/rejected": -258.93450927734375,
|
| 1236 |
+
"loss": 0.6505,
|
| 1237 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1238 |
+
"rewards/chosen": -0.2365461140871048,
|
| 1239 |
+
"rewards/margins": 0.12159979343414307,
|
| 1240 |
+
"rewards/rejected": -0.35814589262008667,
|
| 1241 |
+
"step": 790
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 1.67,
|
| 1245 |
+
"learning_rate": 3.8702478614051353e-07,
|
| 1246 |
+
"logits/chosen": 0.6355623006820679,
|
| 1247 |
+
"logits/rejected": 0.64263516664505,
|
| 1248 |
+
"logps/chosen": -261.3099670410156,
|
| 1249 |
+
"logps/rejected": -265.4583435058594,
|
| 1250 |
+
"loss": 0.6449,
|
| 1251 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 1252 |
+
"rewards/chosen": -0.23573680222034454,
|
| 1253 |
+
"rewards/margins": 0.13916215300559998,
|
| 1254 |
+
"rewards/rejected": -0.3748989403247833,
|
| 1255 |
+
"step": 800
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 1.67,
|
| 1259 |
+
"eval_logits/chosen": 0.6347466111183167,
|
| 1260 |
+
"eval_logits/rejected": 0.7372007966041565,
|
| 1261 |
+
"eval_logps/chosen": -285.4225769042969,
|
| 1262 |
+
"eval_logps/rejected": -268.3813781738281,
|
| 1263 |
+
"eval_loss": 0.6541773676872253,
|
| 1264 |
+
"eval_rewards/accuracies": 0.656000018119812,
|
| 1265 |
+
"eval_rewards/chosen": -0.2517840266227722,
|
| 1266 |
+
"eval_rewards/margins": 0.11766859143972397,
|
| 1267 |
+
"eval_rewards/rejected": -0.3694525957107544,
|
| 1268 |
+
"eval_runtime": 325.7549,
|
| 1269 |
+
"eval_samples_per_second": 6.14,
|
| 1270 |
+
"eval_steps_per_second": 0.384,
|
| 1271 |
+
"step": 800
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 1.7,
|
| 1275 |
+
"learning_rate": 3.3952790595787986e-07,
|
| 1276 |
+
"logits/chosen": 0.7447446584701538,
|
| 1277 |
+
"logits/rejected": 0.7175893783569336,
|
| 1278 |
+
"logps/chosen": -275.09906005859375,
|
| 1279 |
+
"logps/rejected": -265.5423889160156,
|
| 1280 |
+
"loss": 0.6508,
|
| 1281 |
+
"rewards/accuracies": 0.581250011920929,
|
| 1282 |
+
"rewards/chosen": -0.25975173711776733,
|
| 1283 |
+
"rewards/margins": 0.08070627599954605,
|
| 1284 |
+
"rewards/rejected": -0.3404580354690552,
|
| 1285 |
+
"step": 810
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 1.72,
|
| 1289 |
+
"learning_rate": 2.9492720416985004e-07,
|
| 1290 |
+
"logits/chosen": 0.5959675908088684,
|
| 1291 |
+
"logits/rejected": 0.6474281549453735,
|
| 1292 |
+
"logps/chosen": -294.9203186035156,
|
| 1293 |
+
"logps/rejected": -284.7016906738281,
|
| 1294 |
+
"loss": 0.6455,
|
| 1295 |
+
"rewards/accuracies": 0.71875,
|
| 1296 |
+
"rewards/chosen": -0.2046935260295868,
|
| 1297 |
+
"rewards/margins": 0.13686785101890564,
|
| 1298 |
+
"rewards/rejected": -0.34156137704849243,
|
| 1299 |
+
"step": 820
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 1.74,
|
| 1303 |
+
"learning_rate": 2.5328246937043526e-07,
|
| 1304 |
+
"logits/chosen": 0.6916497349739075,
|
| 1305 |
+
"logits/rejected": 0.7139278650283813,
|
| 1306 |
+
"logps/chosen": -264.2368469238281,
|
| 1307 |
+
"logps/rejected": -253.6294403076172,
|
| 1308 |
+
"loss": 0.6506,
|
| 1309 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1310 |
+
"rewards/chosen": -0.28065750002861023,
|
| 1311 |
+
"rewards/margins": 0.10091479122638702,
|
| 1312 |
+
"rewards/rejected": -0.38157230615615845,
|
| 1313 |
+
"step": 830
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 1.76,
|
| 1317 |
+
"learning_rate": 2.1464952759020857e-07,
|
| 1318 |
+
"logits/chosen": 0.6801538467407227,
|
| 1319 |
+
"logits/rejected": 0.7134484648704529,
|
| 1320 |
+
"logps/chosen": -274.87481689453125,
|
| 1321 |
+
"logps/rejected": -282.79766845703125,
|
| 1322 |
+
"loss": 0.6522,
|
| 1323 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1324 |
+
"rewards/chosen": -0.2523757517337799,
|
| 1325 |
+
"rewards/margins": 0.11816386878490448,
|
| 1326 |
+
"rewards/rejected": -0.3705395758152008,
|
| 1327 |
+
"step": 840
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 1.78,
|
| 1331 |
+
"learning_rate": 1.790801674598186e-07,
|
| 1332 |
+
"logits/chosen": 0.6462175846099854,
|
| 1333 |
+
"logits/rejected": 0.6658031940460205,
|
| 1334 |
+
"logps/chosen": -292.3714599609375,
|
| 1335 |
+
"logps/rejected": -269.7221984863281,
|
| 1336 |
+
"loss": 0.6522,
|
| 1337 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1338 |
+
"rewards/chosen": -0.24048607051372528,
|
| 1339 |
+
"rewards/margins": 0.10906896740198135,
|
| 1340 |
+
"rewards/rejected": -0.34955504536628723,
|
| 1341 |
+
"step": 850
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 1.8,
|
| 1345 |
+
"learning_rate": 1.4662207078575685e-07,
|
| 1346 |
+
"logits/chosen": 0.6442946791648865,
|
| 1347 |
+
"logits/rejected": 0.7183449864387512,
|
| 1348 |
+
"logps/chosen": -256.558837890625,
|
| 1349 |
+
"logps/rejected": -274.9573669433594,
|
| 1350 |
+
"loss": 0.6476,
|
| 1351 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1352 |
+
"rewards/chosen": -0.2482720911502838,
|
| 1353 |
+
"rewards/margins": 0.1144469827413559,
|
| 1354 |
+
"rewards/rejected": -0.3627190887928009,
|
| 1355 |
+
"step": 860
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 1.82,
|
| 1359 |
+
"learning_rate": 1.1731874863145143e-07,
|
| 1360 |
+
"logits/chosen": 0.68730229139328,
|
| 1361 |
+
"logits/rejected": 0.7417377829551697,
|
| 1362 |
+
"logps/chosen": -285.8447570800781,
|
| 1363 |
+
"logps/rejected": -258.12408447265625,
|
| 1364 |
+
"loss": 0.6483,
|
| 1365 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1366 |
+
"rewards/chosen": -0.24029162526130676,
|
| 1367 |
+
"rewards/margins": 0.10951662063598633,
|
| 1368 |
+
"rewards/rejected": -0.3498082160949707,
|
| 1369 |
+
"step": 870
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 1.84,
|
| 1373 |
+
"learning_rate": 9.120948298936422e-08,
|
| 1374 |
+
"logits/chosen": 0.7039914131164551,
|
| 1375 |
+
"logits/rejected": 0.720660388469696,
|
| 1376 |
+
"logps/chosen": -275.22564697265625,
|
| 1377 |
+
"logps/rejected": -259.50921630859375,
|
| 1378 |
+
"loss": 0.6391,
|
| 1379 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1380 |
+
"rewards/chosen": -0.2585623860359192,
|
| 1381 |
+
"rewards/margins": 0.10217638313770294,
|
| 1382 |
+
"rewards/rejected": -0.36073875427246094,
|
| 1383 |
+
"step": 880
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 1.86,
|
| 1387 |
+
"learning_rate": 6.832927412229017e-08,
|
| 1388 |
+
"logits/chosen": 0.634792685508728,
|
| 1389 |
+
"logits/rejected": 0.7409111857414246,
|
| 1390 |
+
"logps/chosen": -285.78076171875,
|
| 1391 |
+
"logps/rejected": -263.0660095214844,
|
| 1392 |
+
"loss": 0.6441,
|
| 1393 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1394 |
+
"rewards/chosen": -0.24097540974617004,
|
| 1395 |
+
"rewards/margins": 0.12255308777093887,
|
| 1396 |
+
"rewards/rejected": -0.3635285198688507,
|
| 1397 |
+
"step": 890
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 1.88,
|
| 1401 |
+
"learning_rate": 4.870879364444109e-08,
|
| 1402 |
+
"logits/chosen": 0.6492586731910706,
|
| 1403 |
+
"logits/rejected": 0.7190669775009155,
|
| 1404 |
+
"logps/chosen": -275.6029357910156,
|
| 1405 |
+
"logps/rejected": -243.29281616210938,
|
| 1406 |
+
"loss": 0.6487,
|
| 1407 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1408 |
+
"rewards/chosen": -0.24271516501903534,
|
| 1409 |
+
"rewards/margins": 0.14071747660636902,
|
| 1410 |
+
"rewards/rejected": -0.38343262672424316,
|
| 1411 |
+
"step": 900
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 1.88,
|
| 1415 |
+
"eval_logits/chosen": 0.6298695802688599,
|
| 1416 |
+
"eval_logits/rejected": 0.7320161461830139,
|
| 1417 |
+
"eval_logps/chosen": -285.9532165527344,
|
| 1418 |
+
"eval_logps/rejected": -269.07244873046875,
|
| 1419 |
+
"eval_loss": 0.6538792848587036,
|
| 1420 |
+
"eval_rewards/accuracies": 0.656000018119812,
|
| 1421 |
+
"eval_rewards/chosen": -0.25709065794944763,
|
| 1422 |
+
"eval_rewards/margins": 0.11927253752946854,
|
| 1423 |
+
"eval_rewards/rejected": -0.37636318802833557,
|
| 1424 |
+
"eval_runtime": 326.0646,
|
| 1425 |
+
"eval_samples_per_second": 6.134,
|
| 1426 |
+
"eval_steps_per_second": 0.383,
|
| 1427 |
+
"step": 900
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 1.9,
|
| 1431 |
+
"learning_rate": 3.237434340521789e-08,
|
| 1432 |
+
"logits/chosen": 0.6427167654037476,
|
| 1433 |
+
"logits/rejected": 0.7207670211791992,
|
| 1434 |
+
"logps/chosen": -286.7334289550781,
|
| 1435 |
+
"logps/rejected": -282.4278259277344,
|
| 1436 |
+
"loss": 0.6455,
|
| 1437 |
+
"rewards/accuracies": 0.65625,
|
| 1438 |
+
"rewards/chosen": -0.24073800444602966,
|
| 1439 |
+
"rewards/margins": 0.1223975196480751,
|
| 1440 |
+
"rewards/rejected": -0.36313554644584656,
|
| 1441 |
+
"step": 910
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 1.93,
|
| 1445 |
+
"learning_rate": 1.93478202307823e-08,
|
| 1446 |
+
"logits/chosen": 0.6422888040542603,
|
| 1447 |
+
"logits/rejected": 0.6643663644790649,
|
| 1448 |
+
"logps/chosen": -267.14398193359375,
|
| 1449 |
+
"logps/rejected": -278.8194274902344,
|
| 1450 |
+
"loss": 0.6473,
|
| 1451 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1452 |
+
"rewards/chosen": -0.257813036441803,
|
| 1453 |
+
"rewards/margins": 0.0857832059264183,
|
| 1454 |
+
"rewards/rejected": -0.34359627962112427,
|
| 1455 |
+
"step": 920
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 1.95,
|
| 1459 |
+
"learning_rate": 9.646686570697062e-09,
|
| 1460 |
+
"logits/chosen": 0.6971439123153687,
|
| 1461 |
+
"logits/rejected": 0.722652792930603,
|
| 1462 |
+
"logps/chosen": -283.13067626953125,
|
| 1463 |
+
"logps/rejected": -285.1507568359375,
|
| 1464 |
+
"loss": 0.6572,
|
| 1465 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1466 |
+
"rewards/chosen": -0.2655975818634033,
|
| 1467 |
+
"rewards/margins": 0.10348521173000336,
|
| 1468 |
+
"rewards/rejected": -0.36908283829689026,
|
| 1469 |
+
"step": 930
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 1.97,
|
| 1473 |
+
"learning_rate": 3.283947088983663e-09,
|
| 1474 |
+
"logits/chosen": 0.6804937720298767,
|
| 1475 |
+
"logits/rejected": 0.6877144575119019,
|
| 1476 |
+
"logps/chosen": -265.1544494628906,
|
| 1477 |
+
"logps/rejected": -278.4422302246094,
|
| 1478 |
+
"loss": 0.6544,
|
| 1479 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 1480 |
+
"rewards/chosen": -0.27914106845855713,
|
| 1481 |
+
"rewards/margins": 0.08945213258266449,
|
| 1482 |
+
"rewards/rejected": -0.3685932457447052,
|
| 1483 |
+
"step": 940
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 1.99,
|
| 1487 |
+
"learning_rate": 2.681312309735229e-10,
|
| 1488 |
+
"logits/chosen": 0.6534776091575623,
|
| 1489 |
+
"logits/rejected": 0.7444522976875305,
|
| 1490 |
+
"logps/chosen": -257.7799377441406,
|
| 1491 |
+
"logps/rejected": -263.00030517578125,
|
| 1492 |
+
"loss": 0.654,
|
| 1493 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1494 |
+
"rewards/chosen": -0.2684522867202759,
|
| 1495 |
+
"rewards/margins": 0.08305440098047256,
|
| 1496 |
+
"rewards/rejected": -0.35150665044784546,
|
| 1497 |
+
"step": 950
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 2.0,
|
| 1501 |
+
"step": 954,
|
| 1502 |
+
"total_flos": 0.0,
|
| 1503 |
+
"train_loss": 0.6680307045922589,
|
| 1504 |
+
"train_runtime": 18174.3674,
|
| 1505 |
+
"train_samples_per_second": 3.364,
|
| 1506 |
+
"train_steps_per_second": 0.052
|
| 1507 |
+
}
|
| 1508 |
+
],
|
| 1509 |
+
"logging_steps": 10,
|
| 1510 |
+
"max_steps": 954,
|
| 1511 |
+
"num_input_tokens_seen": 0,
|
| 1512 |
+
"num_train_epochs": 2,
|
| 1513 |
+
"save_steps": 100,
|
| 1514 |
+
"total_flos": 0.0,
|
| 1515 |
+
"train_batch_size": 4,
|
| 1516 |
+
"trial_name": null,
|
| 1517 |
+
"trial_params": null
|
| 1518 |
+
}
|