Model save
Browse files- README.md +84 -0
- adapter_model.safetensors +1 -1
- all_results.json +21 -0
- eval_results.json +16 -0
- train_results.json +8 -0
- trainer_state.json +1972 -0
README.md
ADDED
|
@@ -0,0 +1,84 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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-test-iter-0
|
| 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-test-iter-0
|
| 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: 0.0002
|
| 22 |
+
- Rewards/chosen: -0.0029
|
| 23 |
+
- Rewards/rejected: -0.0032
|
| 24 |
+
- Rewards/accuracies: 0.5130
|
| 25 |
+
- Rewards/margins: 0.0003
|
| 26 |
+
- Logps/rejected: -233.8547
|
| 27 |
+
- Logps/chosen: -256.9005
|
| 28 |
+
- Logits/rejected: 0.8721
|
| 29 |
+
- Logits/chosen: 0.8145
|
| 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 |
+
- gradient_accumulation_steps: 4
|
| 54 |
+
- total_train_batch_size: 16
|
| 55 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 56 |
+
- lr_scheduler_type: cosine
|
| 57 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 58 |
+
- num_epochs: 4
|
| 59 |
+
|
| 60 |
+
### Training results
|
| 61 |
+
|
| 62 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 63 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 64 |
+
| 0.0001 | 0.32 | 100 | 0.0002 | -0.0012 | -0.0015 | 0.5200 | 0.0003 | -233.6874 | -256.7341 | 0.8840 | 0.8263 |
|
| 65 |
+
| 0.0001 | 0.64 | 200 | 0.0002 | -0.0021 | -0.0023 | 0.5005 | 0.0002 | -233.7691 | -256.8278 | 0.8778 | 0.8201 |
|
| 66 |
+
| 0.0001 | 0.96 | 300 | 0.0002 | -0.0021 | -0.0024 | 0.4985 | 0.0003 | -233.7780 | -256.8272 | 0.8783 | 0.8206 |
|
| 67 |
+
| 0.0001 | 1.28 | 400 | 0.0002 | -0.0026 | -0.0029 | 0.5195 | 0.0003 | -233.8277 | -256.8757 | 0.8769 | 0.8192 |
|
| 68 |
+
| 0.0001 | 1.6 | 500 | 0.0002 | -0.0027 | -0.0030 | 0.5170 | 0.0003 | -233.8388 | -256.8869 | 0.8729 | 0.8151 |
|
| 69 |
+
| 0.0001 | 1.92 | 600 | 0.0002 | -0.0027 | -0.0030 | 0.5070 | 0.0003 | -233.8414 | -256.8860 | 0.8757 | 0.8180 |
|
| 70 |
+
| 0.0001 | 2.24 | 700 | 0.0002 | -0.0030 | -0.0032 | 0.5065 | 0.0002 | -233.8592 | -256.9123 | 0.8719 | 0.8142 |
|
| 71 |
+
| 0.0001 | 2.56 | 800 | 0.0002 | -0.0028 | -0.0030 | 0.5190 | 0.0003 | -233.8422 | -256.8898 | 0.8713 | 0.8135 |
|
| 72 |
+
| 0.0001 | 2.88 | 900 | 0.0002 | -0.0030 | -0.0031 | 0.5015 | 0.0002 | -233.8529 | -256.9111 | 0.8714 | 0.8136 |
|
| 73 |
+
| 0.0001 | 3.2 | 1000 | 0.0002 | -0.0029 | -0.0033 | 0.5180 | 0.0004 | -233.8666 | -256.9036 | 0.8733 | 0.8156 |
|
| 74 |
+
| 0.0001 | 3.52 | 1100 | 0.0002 | -0.0029 | -0.0034 | 0.5265 | 0.0005 | -233.8779 | -256.9080 | 0.8724 | 0.8145 |
|
| 75 |
+
| 0.0001 | 3.84 | 1200 | 0.0002 | -0.0031 | -0.0033 | 0.5045 | 0.0003 | -233.8733 | -256.9227 | 0.8705 | 0.8127 |
|
| 76 |
+
|
| 77 |
+
|
| 78 |
+
### Framework versions
|
| 79 |
+
|
| 80 |
+
- PEFT 0.7.1
|
| 81 |
+
- Transformers 4.36.2
|
| 82 |
+
- Pytorch 2.2.1+cu121
|
| 83 |
+
- Datasets 2.14.6
|
| 84 |
+
- 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 sha256:08f46b208abcd902f8b18e4179aea06a78b6e88e8af55a35beb190f9b3ea699c
|
| 3 |
size 41977616
|
all_results.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.99,
|
| 3 |
+
"eval_logits/chosen": 0.81451815366745,
|
| 4 |
+
"eval_logits/rejected": 0.8721050024032593,
|
| 5 |
+
"eval_logps/chosen": -256.90045166015625,
|
| 6 |
+
"eval_logps/rejected": -233.8546905517578,
|
| 7 |
+
"eval_loss": 0.00017304150969721377,
|
| 8 |
+
"eval_rewards/accuracies": 0.5130000114440918,
|
| 9 |
+
"eval_rewards/chosen": -0.0028624406550079584,
|
| 10 |
+
"eval_rewards/margins": 0.0003016398404724896,
|
| 11 |
+
"eval_rewards/rejected": -0.003164080437272787,
|
| 12 |
+
"eval_runtime": 412.2426,
|
| 13 |
+
"eval_samples": 2000,
|
| 14 |
+
"eval_samples_per_second": 4.852,
|
| 15 |
+
"eval_steps_per_second": 1.213,
|
| 16 |
+
"train_loss": 9.35627439654733e-05,
|
| 17 |
+
"train_runtime": 14053.8323,
|
| 18 |
+
"train_samples": 61135,
|
| 19 |
+
"train_samples_per_second": 1.423,
|
| 20 |
+
"train_steps_per_second": 0.089
|
| 21 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.99,
|
| 3 |
+
"eval_logits/chosen": 0.81451815366745,
|
| 4 |
+
"eval_logits/rejected": 0.8721050024032593,
|
| 5 |
+
"eval_logps/chosen": -256.90045166015625,
|
| 6 |
+
"eval_logps/rejected": -233.8546905517578,
|
| 7 |
+
"eval_loss": 0.00017304150969721377,
|
| 8 |
+
"eval_rewards/accuracies": 0.5130000114440918,
|
| 9 |
+
"eval_rewards/chosen": -0.0028624406550079584,
|
| 10 |
+
"eval_rewards/margins": 0.0003016398404724896,
|
| 11 |
+
"eval_rewards/rejected": -0.003164080437272787,
|
| 12 |
+
"eval_runtime": 412.2426,
|
| 13 |
+
"eval_samples": 2000,
|
| 14 |
+
"eval_samples_per_second": 4.852,
|
| 15 |
+
"eval_steps_per_second": 1.213
|
| 16 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.99,
|
| 3 |
+
"train_loss": 9.35627439654733e-05,
|
| 4 |
+
"train_runtime": 14053.8323,
|
| 5 |
+
"train_samples": 61135,
|
| 6 |
+
"train_samples_per_second": 1.423,
|
| 7 |
+
"train_steps_per_second": 0.089
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1972 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.9936,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 1248,
|
| 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": 4e-08,
|
| 14 |
+
"logits/chosen": 0.76749187707901,
|
| 15 |
+
"logits/rejected": 1.0001295804977417,
|
| 16 |
+
"logps/chosen": -205.27383422851562,
|
| 17 |
+
"logps/rejected": -130.56936645507812,
|
| 18 |
+
"loss": 0.0001,
|
| 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": 0.03,
|
| 27 |
+
"learning_rate": 4.0000000000000003e-07,
|
| 28 |
+
"logits/chosen": 0.7141001224517822,
|
| 29 |
+
"logits/rejected": 0.7724499106407166,
|
| 30 |
+
"logps/chosen": -190.74786376953125,
|
| 31 |
+
"logps/rejected": -138.00537109375,
|
| 32 |
+
"loss": 0.0001,
|
| 33 |
+
"rewards/accuracies": 0.4027777910232544,
|
| 34 |
+
"rewards/chosen": 0.0002450596366543323,
|
| 35 |
+
"rewards/margins": 0.0004758307186421007,
|
| 36 |
+
"rewards/rejected": -0.0002307710237801075,
|
| 37 |
+
"step": 10
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.06,
|
| 41 |
+
"learning_rate": 8.000000000000001e-07,
|
| 42 |
+
"logits/chosen": 0.601749062538147,
|
| 43 |
+
"logits/rejected": 0.879185676574707,
|
| 44 |
+
"logps/chosen": -173.01181030273438,
|
| 45 |
+
"logps/rejected": -116.30070495605469,
|
| 46 |
+
"loss": 0.0001,
|
| 47 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 48 |
+
"rewards/chosen": 4.058447757415706e-06,
|
| 49 |
+
"rewards/margins": -0.00013263085565995425,
|
| 50 |
+
"rewards/rejected": 0.00013668931205756962,
|
| 51 |
+
"step": 20
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.1,
|
| 55 |
+
"learning_rate": 1.2000000000000002e-06,
|
| 56 |
+
"logits/chosen": 0.6547126173973083,
|
| 57 |
+
"logits/rejected": 0.833116352558136,
|
| 58 |
+
"logps/chosen": -187.99520874023438,
|
| 59 |
+
"logps/rejected": -127.53055572509766,
|
| 60 |
+
"loss": 0.0001,
|
| 61 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 62 |
+
"rewards/chosen": 2.28220596909523e-06,
|
| 63 |
+
"rewards/margins": 0.00018625140364747494,
|
| 64 |
+
"rewards/rejected": -0.00018396916857454926,
|
| 65 |
+
"step": 30
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.13,
|
| 69 |
+
"learning_rate": 1.6000000000000001e-06,
|
| 70 |
+
"logits/chosen": 0.5628782510757446,
|
| 71 |
+
"logits/rejected": 0.745602548122406,
|
| 72 |
+
"logps/chosen": -183.92758178710938,
|
| 73 |
+
"logps/rejected": -134.64688110351562,
|
| 74 |
+
"loss": 0.0001,
|
| 75 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 76 |
+
"rewards/chosen": -0.0005207593785598874,
|
| 77 |
+
"rewards/margins": -0.0005336635513231158,
|
| 78 |
+
"rewards/rejected": 1.290418458665954e-05,
|
| 79 |
+
"step": 40
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.16,
|
| 83 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 84 |
+
"logits/chosen": 0.6242247819900513,
|
| 85 |
+
"logits/rejected": 0.8314679265022278,
|
| 86 |
+
"logps/chosen": -171.67556762695312,
|
| 87 |
+
"logps/rejected": -120.4903335571289,
|
| 88 |
+
"loss": 0.0001,
|
| 89 |
+
"rewards/accuracies": 0.5,
|
| 90 |
+
"rewards/chosen": 0.00020891983876936138,
|
| 91 |
+
"rewards/margins": 0.0005679951282218099,
|
| 92 |
+
"rewards/rejected": -0.00035907537676393986,
|
| 93 |
+
"step": 50
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.19,
|
| 97 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 98 |
+
"logits/chosen": 0.7160875797271729,
|
| 99 |
+
"logits/rejected": 0.8556329607963562,
|
| 100 |
+
"logps/chosen": -185.2340087890625,
|
| 101 |
+
"logps/rejected": -124.66600036621094,
|
| 102 |
+
"loss": 0.0001,
|
| 103 |
+
"rewards/accuracies": 0.4312500059604645,
|
| 104 |
+
"rewards/chosen": 0.0006271885358728468,
|
| 105 |
+
"rewards/margins": 0.0005722854984924197,
|
| 106 |
+
"rewards/rejected": 5.4903095588088036e-05,
|
| 107 |
+
"step": 60
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.22,
|
| 111 |
+
"learning_rate": 2.8000000000000003e-06,
|
| 112 |
+
"logits/chosen": 0.6916844844818115,
|
| 113 |
+
"logits/rejected": 0.8196538090705872,
|
| 114 |
+
"logps/chosen": -161.62498474121094,
|
| 115 |
+
"logps/rejected": -107.52471923828125,
|
| 116 |
+
"loss": 0.0001,
|
| 117 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 118 |
+
"rewards/chosen": 0.0007440428016707301,
|
| 119 |
+
"rewards/margins": 0.0006342666456475854,
|
| 120 |
+
"rewards/rejected": 0.00010977611964335665,
|
| 121 |
+
"step": 70
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.26,
|
| 125 |
+
"learning_rate": 3.2000000000000003e-06,
|
| 126 |
+
"logits/chosen": 0.6701158285140991,
|
| 127 |
+
"logits/rejected": 0.8293789029121399,
|
| 128 |
+
"logps/chosen": -177.7081298828125,
|
| 129 |
+
"logps/rejected": -122.1727294921875,
|
| 130 |
+
"loss": 0.0001,
|
| 131 |
+
"rewards/accuracies": 0.512499988079071,
|
| 132 |
+
"rewards/chosen": 0.0014677448198199272,
|
| 133 |
+
"rewards/margins": 0.000520342611707747,
|
| 134 |
+
"rewards/rejected": 0.0009474022081121802,
|
| 135 |
+
"step": 80
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.29,
|
| 139 |
+
"learning_rate": 3.6000000000000003e-06,
|
| 140 |
+
"logits/chosen": 0.6228800415992737,
|
| 141 |
+
"logits/rejected": 0.7212022542953491,
|
| 142 |
+
"logps/chosen": -183.03811645507812,
|
| 143 |
+
"logps/rejected": -139.0586700439453,
|
| 144 |
+
"loss": 0.0001,
|
| 145 |
+
"rewards/accuracies": 0.53125,
|
| 146 |
+
"rewards/chosen": 0.0022542846854776144,
|
| 147 |
+
"rewards/margins": 0.0009365282021462917,
|
| 148 |
+
"rewards/rejected": 0.0013177564833313227,
|
| 149 |
+
"step": 90
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.32,
|
| 153 |
+
"learning_rate": 4.000000000000001e-06,
|
| 154 |
+
"logits/chosen": 0.6344588994979858,
|
| 155 |
+
"logits/rejected": 0.8314794301986694,
|
| 156 |
+
"logps/chosen": -163.23043823242188,
|
| 157 |
+
"logps/rejected": -121.80052185058594,
|
| 158 |
+
"loss": 0.0001,
|
| 159 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 160 |
+
"rewards/chosen": 0.002225330099463463,
|
| 161 |
+
"rewards/margins": 0.0011866830755025148,
|
| 162 |
+
"rewards/rejected": 0.0010386471403762698,
|
| 163 |
+
"step": 100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.32,
|
| 167 |
+
"eval_logits/chosen": 0.8262824416160583,
|
| 168 |
+
"eval_logits/rejected": 0.8840107321739197,
|
| 169 |
+
"eval_logps/chosen": -256.7341003417969,
|
| 170 |
+
"eval_logps/rejected": -233.68739318847656,
|
| 171 |
+
"eval_loss": 0.00015139963943511248,
|
| 172 |
+
"eval_rewards/accuracies": 0.5199999809265137,
|
| 173 |
+
"eval_rewards/chosen": -0.00119913334492594,
|
| 174 |
+
"eval_rewards/margins": 0.0002922326675616205,
|
| 175 |
+
"eval_rewards/rejected": -0.0014913661871105433,
|
| 176 |
+
"eval_runtime": 415.0818,
|
| 177 |
+
"eval_samples_per_second": 4.818,
|
| 178 |
+
"eval_steps_per_second": 1.205,
|
| 179 |
+
"step": 100
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.35,
|
| 183 |
+
"learning_rate": 4.4e-06,
|
| 184 |
+
"logits/chosen": 0.6864518523216248,
|
| 185 |
+
"logits/rejected": 0.9017802476882935,
|
| 186 |
+
"logps/chosen": -174.11480712890625,
|
| 187 |
+
"logps/rejected": -118.93388366699219,
|
| 188 |
+
"loss": 0.0001,
|
| 189 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 190 |
+
"rewards/chosen": 0.00283862859942019,
|
| 191 |
+
"rewards/margins": 0.0018309386214241385,
|
| 192 |
+
"rewards/rejected": 0.001007690210826695,
|
| 193 |
+
"step": 110
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.38,
|
| 197 |
+
"learning_rate": 4.800000000000001e-06,
|
| 198 |
+
"logits/chosen": 0.5682826638221741,
|
| 199 |
+
"logits/rejected": 0.6713820695877075,
|
| 200 |
+
"logps/chosen": -189.4928436279297,
|
| 201 |
+
"logps/rejected": -135.20498657226562,
|
| 202 |
+
"loss": 0.0001,
|
| 203 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 204 |
+
"rewards/chosen": 0.0033208150416612625,
|
| 205 |
+
"rewards/margins": 0.001955473329871893,
|
| 206 |
+
"rewards/rejected": 0.0013653415953740478,
|
| 207 |
+
"step": 120
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.42,
|
| 211 |
+
"learning_rate": 4.999755441268144e-06,
|
| 212 |
+
"logits/chosen": 0.7162402868270874,
|
| 213 |
+
"logits/rejected": 0.8464711904525757,
|
| 214 |
+
"logps/chosen": -175.83523559570312,
|
| 215 |
+
"logps/rejected": -126.2065200805664,
|
| 216 |
+
"loss": 0.0001,
|
| 217 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 218 |
+
"rewards/chosen": 0.0035425268579274416,
|
| 219 |
+
"rewards/margins": 0.0022287473548203707,
|
| 220 |
+
"rewards/rejected": 0.0013137792702764273,
|
| 221 |
+
"step": 130
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.45,
|
| 225 |
+
"learning_rate": 4.997799258487003e-06,
|
| 226 |
+
"logits/chosen": 0.6711681485176086,
|
| 227 |
+
"logits/rejected": 0.8102658987045288,
|
| 228 |
+
"logps/chosen": -182.16258239746094,
|
| 229 |
+
"logps/rejected": -128.60398864746094,
|
| 230 |
+
"loss": 0.0001,
|
| 231 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 232 |
+
"rewards/chosen": 0.0029964938294142485,
|
| 233 |
+
"rewards/margins": 0.002129464875906706,
|
| 234 |
+
"rewards/rejected": 0.0008670290117152035,
|
| 235 |
+
"step": 140
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.48,
|
| 239 |
+
"learning_rate": 4.993888423734898e-06,
|
| 240 |
+
"logits/chosen": 0.6097667217254639,
|
| 241 |
+
"logits/rejected": 0.7144483327865601,
|
| 242 |
+
"logps/chosen": -190.95156860351562,
|
| 243 |
+
"logps/rejected": -141.19229125976562,
|
| 244 |
+
"loss": 0.0001,
|
| 245 |
+
"rewards/accuracies": 0.59375,
|
| 246 |
+
"rewards/chosen": 0.0037058107554912567,
|
| 247 |
+
"rewards/margins": 0.002363733481615782,
|
| 248 |
+
"rewards/rejected": 0.001342077157460153,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.51,
|
| 253 |
+
"learning_rate": 4.988025997434253e-06,
|
| 254 |
+
"logits/chosen": 0.6315719485282898,
|
| 255 |
+
"logits/rejected": 0.7987428307533264,
|
| 256 |
+
"logps/chosen": -181.68386840820312,
|
| 257 |
+
"logps/rejected": -128.19888305664062,
|
| 258 |
+
"loss": 0.0001,
|
| 259 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 260 |
+
"rewards/chosen": 0.002868856769055128,
|
| 261 |
+
"rewards/margins": 0.0025774172972887754,
|
| 262 |
+
"rewards/rejected": 0.00029143941355869174,
|
| 263 |
+
"step": 160
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.54,
|
| 267 |
+
"learning_rate": 4.980216567224801e-06,
|
| 268 |
+
"logits/chosen": 0.6035071611404419,
|
| 269 |
+
"logits/rejected": 0.8083668947219849,
|
| 270 |
+
"logps/chosen": -181.84107971191406,
|
| 271 |
+
"logps/rejected": -126.97474670410156,
|
| 272 |
+
"loss": 0.0001,
|
| 273 |
+
"rewards/accuracies": 0.612500011920929,
|
| 274 |
+
"rewards/chosen": 0.0037610617000609636,
|
| 275 |
+
"rewards/margins": 0.0028130013961344957,
|
| 276 |
+
"rewards/rejected": 0.0009480599546805024,
|
| 277 |
+
"step": 170
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.58,
|
| 281 |
+
"learning_rate": 4.970466244373527e-06,
|
| 282 |
+
"logits/chosen": 0.6453949213027954,
|
| 283 |
+
"logits/rejected": 0.797042727470398,
|
| 284 |
+
"logps/chosen": -155.79519653320312,
|
| 285 |
+
"logps/rejected": -113.65773010253906,
|
| 286 |
+
"loss": 0.0001,
|
| 287 |
+
"rewards/accuracies": 0.574999988079071,
|
| 288 |
+
"rewards/chosen": 0.0029109427705407143,
|
| 289 |
+
"rewards/margins": 0.0023111128248274326,
|
| 290 |
+
"rewards/rejected": 0.0005998298292979598,
|
| 291 |
+
"step": 180
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.61,
|
| 295 |
+
"learning_rate": 4.958782658992307e-06,
|
| 296 |
+
"logits/chosen": 0.6763242483139038,
|
| 297 |
+
"logits/rejected": 0.8502823710441589,
|
| 298 |
+
"logps/chosen": -159.13275146484375,
|
| 299 |
+
"logps/rejected": -120.99656677246094,
|
| 300 |
+
"loss": 0.0001,
|
| 301 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 302 |
+
"rewards/chosen": 0.0026291562244296074,
|
| 303 |
+
"rewards/margins": 0.0016078405315056443,
|
| 304 |
+
"rewards/rejected": 0.001021315692923963,
|
| 305 |
+
"step": 190
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.64,
|
| 309 |
+
"learning_rate": 4.945174954066957e-06,
|
| 310 |
+
"logits/chosen": 0.7041156888008118,
|
| 311 |
+
"logits/rejected": 0.7947753667831421,
|
| 312 |
+
"logps/chosen": -177.0174560546875,
|
| 313 |
+
"logps/rejected": -127.36385345458984,
|
| 314 |
+
"loss": 0.0001,
|
| 315 |
+
"rewards/accuracies": 0.625,
|
| 316 |
+
"rewards/chosen": 0.003798459889367223,
|
| 317 |
+
"rewards/margins": 0.0031563788652420044,
|
| 318 |
+
"rewards/rejected": 0.0006420810823328793,
|
| 319 |
+
"step": 200
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.64,
|
| 323 |
+
"eval_logits/chosen": 0.8200604319572449,
|
| 324 |
+
"eval_logits/rejected": 0.8777603507041931,
|
| 325 |
+
"eval_logps/chosen": -256.8277893066406,
|
| 326 |
+
"eval_logps/rejected": -233.76914978027344,
|
| 327 |
+
"eval_loss": 0.000158556635142304,
|
| 328 |
+
"eval_rewards/accuracies": 0.5005000233650208,
|
| 329 |
+
"eval_rewards/chosen": -0.00213597621768713,
|
| 330 |
+
"eval_rewards/margins": 0.00017285306239500642,
|
| 331 |
+
"eval_rewards/rejected": -0.0023088292218744755,
|
| 332 |
+
"eval_runtime": 412.468,
|
| 333 |
+
"eval_samples_per_second": 4.849,
|
| 334 |
+
"eval_steps_per_second": 1.212,
|
| 335 |
+
"step": 200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 0.67,
|
| 339 |
+
"learning_rate": 4.929653778302397e-06,
|
| 340 |
+
"logits/chosen": 0.6099511981010437,
|
| 341 |
+
"logits/rejected": 0.811872661113739,
|
| 342 |
+
"logps/chosen": -192.26446533203125,
|
| 343 |
+
"logps/rejected": -131.25962829589844,
|
| 344 |
+
"loss": 0.0001,
|
| 345 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 346 |
+
"rewards/chosen": 0.0029747250955551863,
|
| 347 |
+
"rewards/margins": 0.0029428431298583746,
|
| 348 |
+
"rewards/rejected": 3.1882118491921574e-05,
|
| 349 |
+
"step": 210
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 0.7,
|
| 353 |
+
"learning_rate": 4.912231277789509e-06,
|
| 354 |
+
"logits/chosen": 0.6272088289260864,
|
| 355 |
+
"logits/rejected": 0.767717719078064,
|
| 356 |
+
"logps/chosen": -193.66490173339844,
|
| 357 |
+
"logps/rejected": -143.585205078125,
|
| 358 |
+
"loss": 0.0001,
|
| 359 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 360 |
+
"rewards/chosen": 0.003646609140560031,
|
| 361 |
+
"rewards/margins": 0.0021339866798371077,
|
| 362 |
+
"rewards/rejected": 0.0015126224607229233,
|
| 363 |
+
"step": 220
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.74,
|
| 367 |
+
"learning_rate": 4.892921086500219e-06,
|
| 368 |
+
"logits/chosen": 0.6231340169906616,
|
| 369 |
+
"logits/rejected": 0.7231715321540833,
|
| 370 |
+
"logps/chosen": -160.1138458251953,
|
| 371 |
+
"logps/rejected": -114.69986724853516,
|
| 372 |
+
"loss": 0.0001,
|
| 373 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 374 |
+
"rewards/chosen": 0.003955821972340345,
|
| 375 |
+
"rewards/margins": 0.003026761580258608,
|
| 376 |
+
"rewards/rejected": 0.0009290605084970593,
|
| 377 |
+
"step": 230
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.77,
|
| 381 |
+
"learning_rate": 4.871738315618236e-06,
|
| 382 |
+
"logits/chosen": 0.7013573050498962,
|
| 383 |
+
"logits/rejected": 0.8051062822341919,
|
| 384 |
+
"logps/chosen": -186.76821899414062,
|
| 385 |
+
"logps/rejected": -135.32489013671875,
|
| 386 |
+
"loss": 0.0001,
|
| 387 |
+
"rewards/accuracies": 0.59375,
|
| 388 |
+
"rewards/chosen": 0.003390461904928088,
|
| 389 |
+
"rewards/margins": 0.002614812459796667,
|
| 390 |
+
"rewards/rejected": 0.0007756495615467429,
|
| 391 |
+
"step": 240
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.8,
|
| 395 |
+
"learning_rate": 4.848699541713801e-06,
|
| 396 |
+
"logits/chosen": 0.6483660340309143,
|
| 397 |
+
"logits/rejected": 0.8247146606445312,
|
| 398 |
+
"logps/chosen": -159.59156799316406,
|
| 399 |
+
"logps/rejected": -118.78582763671875,
|
| 400 |
+
"loss": 0.0001,
|
| 401 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 402 |
+
"rewards/chosen": 0.0031985179521143436,
|
| 403 |
+
"rewards/margins": 0.0021302015520632267,
|
| 404 |
+
"rewards/rejected": 0.0010683165164664388,
|
| 405 |
+
"step": 250
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.83,
|
| 409 |
+
"learning_rate": 4.823822793771696e-06,
|
| 410 |
+
"logits/chosen": 0.625135064125061,
|
| 411 |
+
"logits/rejected": 0.7819581627845764,
|
| 412 |
+
"logps/chosen": -169.98837280273438,
|
| 413 |
+
"logps/rejected": -119.9397964477539,
|
| 414 |
+
"loss": 0.0001,
|
| 415 |
+
"rewards/accuracies": 0.574999988079071,
|
| 416 |
+
"rewards/chosen": 0.00324636441655457,
|
| 417 |
+
"rewards/margins": 0.0028731045313179493,
|
| 418 |
+
"rewards/rejected": 0.0003732596233021468,
|
| 419 |
+
"step": 260
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 0.86,
|
| 423 |
+
"learning_rate": 4.797127539082669e-06,
|
| 424 |
+
"logits/chosen": 0.649788498878479,
|
| 425 |
+
"logits/rejected": 0.8432229161262512,
|
| 426 |
+
"logps/chosen": -191.59213256835938,
|
| 427 |
+
"logps/rejected": -134.23428344726562,
|
| 428 |
+
"loss": 0.0001,
|
| 429 |
+
"rewards/accuracies": 0.574999988079071,
|
| 430 |
+
"rewards/chosen": 0.0034480884205549955,
|
| 431 |
+
"rewards/margins": 0.002935568569228053,
|
| 432 |
+
"rewards/rejected": 0.0005125202005729079,
|
| 433 |
+
"step": 270
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.9,
|
| 437 |
+
"learning_rate": 4.7686346680093135e-06,
|
| 438 |
+
"logits/chosen": 0.6038953065872192,
|
| 439 |
+
"logits/rejected": 0.8061636686325073,
|
| 440 |
+
"logps/chosen": -188.86422729492188,
|
| 441 |
+
"logps/rejected": -125.95369720458984,
|
| 442 |
+
"loss": 0.0001,
|
| 443 |
+
"rewards/accuracies": 0.71875,
|
| 444 |
+
"rewards/chosen": 0.004464815836399794,
|
| 445 |
+
"rewards/margins": 0.004056466277688742,
|
| 446 |
+
"rewards/rejected": 0.0004083492676727474,
|
| 447 |
+
"step": 280
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.93,
|
| 451 |
+
"learning_rate": 4.738366477638319e-06,
|
| 452 |
+
"logits/chosen": 0.5706946849822998,
|
| 453 |
+
"logits/rejected": 0.7165879011154175,
|
| 454 |
+
"logps/chosen": -172.26846313476562,
|
| 455 |
+
"logps/rejected": -113.09400939941406,
|
| 456 |
+
"loss": 0.0001,
|
| 457 |
+
"rewards/accuracies": 0.637499988079071,
|
| 458 |
+
"rewards/chosen": 0.0033409663010388613,
|
| 459 |
+
"rewards/margins": 0.0031195811461657286,
|
| 460 |
+
"rewards/rejected": 0.0002213849511463195,
|
| 461 |
+
"step": 290
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.96,
|
| 465 |
+
"learning_rate": 4.7063466543318965e-06,
|
| 466 |
+
"logits/chosen": 0.6467469334602356,
|
| 467 |
+
"logits/rejected": 0.8215667605400085,
|
| 468 |
+
"logps/chosen": -171.28201293945312,
|
| 469 |
+
"logps/rejected": -121.31685638427734,
|
| 470 |
+
"loss": 0.0001,
|
| 471 |
+
"rewards/accuracies": 0.606249988079071,
|
| 472 |
+
"rewards/chosen": 0.00259224371984601,
|
| 473 |
+
"rewards/margins": 0.0022495179437100887,
|
| 474 |
+
"rewards/rejected": 0.0003427262417972088,
|
| 475 |
+
"step": 300
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.96,
|
| 479 |
+
"eval_logits/chosen": 0.8206124901771545,
|
| 480 |
+
"eval_logits/rejected": 0.8782824277877808,
|
| 481 |
+
"eval_logps/chosen": -256.82720947265625,
|
| 482 |
+
"eval_logps/rejected": -233.77804565429688,
|
| 483 |
+
"eval_loss": 0.00015650840941816568,
|
| 484 |
+
"eval_rewards/accuracies": 0.4984999895095825,
|
| 485 |
+
"eval_rewards/chosen": -0.002130264649167657,
|
| 486 |
+
"eval_rewards/margins": 0.00026750334654934704,
|
| 487 |
+
"eval_rewards/rejected": -0.0023977679666131735,
|
| 488 |
+
"eval_runtime": 412.0828,
|
| 489 |
+
"eval_samples_per_second": 4.853,
|
| 490 |
+
"eval_steps_per_second": 1.213,
|
| 491 |
+
"step": 300
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.99,
|
| 495 |
+
"learning_rate": 4.672600255192022e-06,
|
| 496 |
+
"logits/chosen": 0.6745213270187378,
|
| 497 |
+
"logits/rejected": 0.7582186460494995,
|
| 498 |
+
"logps/chosen": -177.1463623046875,
|
| 499 |
+
"logps/rejected": -121.45220947265625,
|
| 500 |
+
"loss": 0.0001,
|
| 501 |
+
"rewards/accuracies": 0.59375,
|
| 502 |
+
"rewards/chosen": 0.003095152322202921,
|
| 503 |
+
"rewards/margins": 0.002747387159615755,
|
| 504 |
+
"rewards/rejected": 0.00034776475513353944,
|
| 505 |
+
"step": 310
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 1.02,
|
| 509 |
+
"learning_rate": 4.6371536884520115e-06,
|
| 510 |
+
"logits/chosen": 0.6856507062911987,
|
| 511 |
+
"logits/rejected": 0.8083009719848633,
|
| 512 |
+
"logps/chosen": -175.43466186523438,
|
| 513 |
+
"logps/rejected": -113.60682678222656,
|
| 514 |
+
"loss": 0.0001,
|
| 515 |
+
"rewards/accuracies": 0.5625,
|
| 516 |
+
"rewards/chosen": 0.0029540827963501215,
|
| 517 |
+
"rewards/margins": 0.0019870868418365717,
|
| 518 |
+
"rewards/rejected": 0.0009669959545135498,
|
| 519 |
+
"step": 320
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 1.06,
|
| 523 |
+
"learning_rate": 4.600034692810764e-06,
|
| 524 |
+
"logits/chosen": 0.6629844307899475,
|
| 525 |
+
"logits/rejected": 0.831767737865448,
|
| 526 |
+
"logps/chosen": -181.02310180664062,
|
| 527 |
+
"logps/rejected": -119.25450134277344,
|
| 528 |
+
"loss": 0.0001,
|
| 529 |
+
"rewards/accuracies": 0.643750011920929,
|
| 530 |
+
"rewards/chosen": 0.0041047511622309685,
|
| 531 |
+
"rewards/margins": 0.003629709128290415,
|
| 532 |
+
"rewards/rejected": 0.0004750423540826887,
|
| 533 |
+
"step": 330
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 1.09,
|
| 537 |
+
"learning_rate": 4.561272315725852e-06,
|
| 538 |
+
"logits/chosen": 0.6184499263763428,
|
| 539 |
+
"logits/rejected": 0.7539141178131104,
|
| 540 |
+
"logps/chosen": -186.89682006835938,
|
| 541 |
+
"logps/rejected": -139.5793914794922,
|
| 542 |
+
"loss": 0.0001,
|
| 543 |
+
"rewards/accuracies": 0.606249988079071,
|
| 544 |
+
"rewards/chosen": 0.0036196750588715076,
|
| 545 |
+
"rewards/margins": 0.0029566895682364702,
|
| 546 |
+
"rewards/rejected": 0.0006629853160120547,
|
| 547 |
+
"step": 340
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 1.12,
|
| 551 |
+
"learning_rate": 4.520896890682449e-06,
|
| 552 |
+
"logits/chosen": 0.6621273159980774,
|
| 553 |
+
"logits/rejected": 0.7836776971817017,
|
| 554 |
+
"logps/chosen": -179.44253540039062,
|
| 555 |
+
"logps/rejected": -126.19744873046875,
|
| 556 |
+
"loss": 0.0001,
|
| 557 |
+
"rewards/accuracies": 0.612500011920929,
|
| 558 |
+
"rewards/chosen": 0.003894126508384943,
|
| 559 |
+
"rewards/margins": 0.003107634838670492,
|
| 560 |
+
"rewards/rejected": 0.0007864916697144508,
|
| 561 |
+
"step": 350
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 1.15,
|
| 565 |
+
"learning_rate": 4.478940013455864e-06,
|
| 566 |
+
"logits/chosen": 0.6130011677742004,
|
| 567 |
+
"logits/rejected": 0.7723643183708191,
|
| 568 |
+
"logps/chosen": -179.69866943359375,
|
| 569 |
+
"logps/rejected": -119.17742919921875,
|
| 570 |
+
"loss": 0.0001,
|
| 571 |
+
"rewards/accuracies": 0.675000011920929,
|
| 572 |
+
"rewards/chosen": 0.0030810669995844364,
|
| 573 |
+
"rewards/margins": 0.0031474686693400145,
|
| 574 |
+
"rewards/rejected": -6.640238279942423e-05,
|
| 575 |
+
"step": 360
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 1.18,
|
| 579 |
+
"learning_rate": 4.435434517386281e-06,
|
| 580 |
+
"logits/chosen": 0.6930996179580688,
|
| 581 |
+
"logits/rejected": 0.8360698819160461,
|
| 582 |
+
"logps/chosen": -189.090087890625,
|
| 583 |
+
"logps/rejected": -135.1619873046875,
|
| 584 |
+
"loss": 0.0001,
|
| 585 |
+
"rewards/accuracies": 0.612500011920929,
|
| 586 |
+
"rewards/chosen": 0.003456123173236847,
|
| 587 |
+
"rewards/margins": 0.0029000742360949516,
|
| 588 |
+
"rewards/rejected": 0.0005560485878959298,
|
| 589 |
+
"step": 370
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 1.22,
|
| 593 |
+
"learning_rate": 4.39041444768504e-06,
|
| 594 |
+
"logits/chosen": 0.6620668172836304,
|
| 595 |
+
"logits/rejected": 0.853411853313446,
|
| 596 |
+
"logps/chosen": -177.2210235595703,
|
| 597 |
+
"logps/rejected": -124.18900299072266,
|
| 598 |
+
"loss": 0.0001,
|
| 599 |
+
"rewards/accuracies": 0.643750011920929,
|
| 600 |
+
"rewards/chosen": 0.00355343846604228,
|
| 601 |
+
"rewards/margins": 0.003507003653794527,
|
| 602 |
+
"rewards/rejected": 4.643518332159147e-05,
|
| 603 |
+
"step": 380
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.25,
|
| 607 |
+
"learning_rate": 4.343915034792569e-06,
|
| 608 |
+
"logits/chosen": 0.629216730594635,
|
| 609 |
+
"logits/rejected": 0.7686316967010498,
|
| 610 |
+
"logps/chosen": -184.9043426513672,
|
| 611 |
+
"logps/rejected": -130.17727661132812,
|
| 612 |
+
"loss": 0.0001,
|
| 613 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 614 |
+
"rewards/chosen": 0.003557362360879779,
|
| 615 |
+
"rewards/margins": 0.003719041822478175,
|
| 616 |
+
"rewards/rejected": -0.00016167931607924402,
|
| 617 |
+
"step": 390
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 1.28,
|
| 621 |
+
"learning_rate": 4.295972666808811e-06,
|
| 622 |
+
"logits/chosen": 0.6529077291488647,
|
| 623 |
+
"logits/rejected": 0.8112252354621887,
|
| 624 |
+
"logps/chosen": -172.85665893554688,
|
| 625 |
+
"logps/rejected": -127.3191146850586,
|
| 626 |
+
"loss": 0.0001,
|
| 627 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 628 |
+
"rewards/chosen": 0.0037923946511000395,
|
| 629 |
+
"rewards/margins": 0.003340603783726692,
|
| 630 |
+
"rewards/rejected": 0.00045179054723121226,
|
| 631 |
+
"step": 400
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 1.28,
|
| 635 |
+
"eval_logits/chosen": 0.8192203044891357,
|
| 636 |
+
"eval_logits/rejected": 0.8769342303276062,
|
| 637 |
+
"eval_logps/chosen": -256.87567138671875,
|
| 638 |
+
"eval_logps/rejected": -233.8277130126953,
|
| 639 |
+
"eval_loss": 0.00015775366046000272,
|
| 640 |
+
"eval_rewards/accuracies": 0.5195000171661377,
|
| 641 |
+
"eval_rewards/chosen": -0.002614969853311777,
|
| 642 |
+
"eval_rewards/margins": 0.00027956385747529566,
|
| 643 |
+
"eval_rewards/rejected": -0.002894533798098564,
|
| 644 |
+
"eval_runtime": 412.2856,
|
| 645 |
+
"eval_samples_per_second": 4.851,
|
| 646 |
+
"eval_steps_per_second": 1.213,
|
| 647 |
+
"step": 400
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 1.31,
|
| 651 |
+
"learning_rate": 4.246624861017732e-06,
|
| 652 |
+
"logits/chosen": 0.625112771987915,
|
| 653 |
+
"logits/rejected": 0.7611836791038513,
|
| 654 |
+
"logps/chosen": -178.12124633789062,
|
| 655 |
+
"logps/rejected": -135.6422882080078,
|
| 656 |
+
"loss": 0.0001,
|
| 657 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 658 |
+
"rewards/chosen": 0.004022772889584303,
|
| 659 |
+
"rewards/margins": 0.003282977268099785,
|
| 660 |
+
"rewards/rejected": 0.0007397954468615353,
|
| 661 |
+
"step": 410
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 1.34,
|
| 665 |
+
"learning_rate": 4.195910234528186e-06,
|
| 666 |
+
"logits/chosen": 0.6460477709770203,
|
| 667 |
+
"logits/rejected": 0.7641812562942505,
|
| 668 |
+
"logps/chosen": -168.17930603027344,
|
| 669 |
+
"logps/rejected": -124.3600082397461,
|
| 670 |
+
"loss": 0.0001,
|
| 671 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 672 |
+
"rewards/chosen": 0.0026535852812230587,
|
| 673 |
+
"rewards/margins": 0.0026994948275387287,
|
| 674 |
+
"rewards/rejected": -4.590977187035605e-05,
|
| 675 |
+
"step": 420
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.38,
|
| 679 |
+
"learning_rate": 4.143868474054098e-06,
|
| 680 |
+
"logits/chosen": 0.5987478494644165,
|
| 681 |
+
"logits/rejected": 0.7457458972930908,
|
| 682 |
+
"logps/chosen": -189.61422729492188,
|
| 683 |
+
"logps/rejected": -123.73170471191406,
|
| 684 |
+
"loss": 0.0001,
|
| 685 |
+
"rewards/accuracies": 0.643750011920929,
|
| 686 |
+
"rewards/chosen": 0.003896042238920927,
|
| 687 |
+
"rewards/margins": 0.00355791044421494,
|
| 688 |
+
"rewards/rejected": 0.00033813173649832606,
|
| 689 |
+
"step": 430
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 1.41,
|
| 693 |
+
"learning_rate": 4.0905403048576545e-06,
|
| 694 |
+
"logits/chosen": 0.5944562554359436,
|
| 695 |
+
"logits/rejected": 0.7794451713562012,
|
| 696 |
+
"logps/chosen": -173.83494567871094,
|
| 697 |
+
"logps/rejected": -125.55583190917969,
|
| 698 |
+
"loss": 0.0001,
|
| 699 |
+
"rewards/accuracies": 0.581250011920929,
|
| 700 |
+
"rewards/chosen": 0.003923391457647085,
|
| 701 |
+
"rewards/margins": 0.003630859311670065,
|
| 702 |
+
"rewards/rejected": 0.00029253208776935935,
|
| 703 |
+
"step": 440
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 1.44,
|
| 707 |
+
"learning_rate": 4.035967458879751e-06,
|
| 708 |
+
"logits/chosen": 0.5830662846565247,
|
| 709 |
+
"logits/rejected": 0.7792474031448364,
|
| 710 |
+
"logps/chosen": -174.3615264892578,
|
| 711 |
+
"logps/rejected": -130.67514038085938,
|
| 712 |
+
"loss": 0.0001,
|
| 713 |
+
"rewards/accuracies": 0.643750011920929,
|
| 714 |
+
"rewards/chosen": 0.0029520168900489807,
|
| 715 |
+
"rewards/margins": 0.0029378398321568966,
|
| 716 |
+
"rewards/rejected": 1.4177237972035073e-05,
|
| 717 |
+
"step": 450
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 1.47,
|
| 721 |
+
"learning_rate": 3.980192642082682e-06,
|
| 722 |
+
"logits/chosen": 0.6810758709907532,
|
| 723 |
+
"logits/rejected": 0.7834008932113647,
|
| 724 |
+
"logps/chosen": -168.80197143554688,
|
| 725 |
+
"logps/rejected": -118.42387390136719,
|
| 726 |
+
"loss": 0.0001,
|
| 727 |
+
"rewards/accuracies": 0.581250011920929,
|
| 728 |
+
"rewards/chosen": 0.0025485253427177668,
|
| 729 |
+
"rewards/margins": 0.002175524365156889,
|
| 730 |
+
"rewards/rejected": 0.00037300080293789506,
|
| 731 |
+
"step": 460
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 1.5,
|
| 735 |
+
"learning_rate": 3.923259501030604e-06,
|
| 736 |
+
"logits/chosen": 0.6081482768058777,
|
| 737 |
+
"logits/rejected": 0.8008432388305664,
|
| 738 |
+
"logps/chosen": -158.7695770263672,
|
| 739 |
+
"logps/rejected": -117.162109375,
|
| 740 |
+
"loss": 0.0001,
|
| 741 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 742 |
+
"rewards/chosen": 0.003074315609410405,
|
| 743 |
+
"rewards/margins": 0.0024462412111461163,
|
| 744 |
+
"rewards/rejected": 0.0006280745146796107,
|
| 745 |
+
"step": 470
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 1.54,
|
| 749 |
+
"learning_rate": 3.865212588733927e-06,
|
| 750 |
+
"logits/chosen": 0.5983234643936157,
|
| 751 |
+
"logits/rejected": 0.8347161412239075,
|
| 752 |
+
"logps/chosen": -177.07232666015625,
|
| 753 |
+
"logps/rejected": -113.17292785644531,
|
| 754 |
+
"loss": 0.0001,
|
| 755 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 756 |
+
"rewards/chosen": 0.004156365990638733,
|
| 757 |
+
"rewards/margins": 0.0032986297737807035,
|
| 758 |
+
"rewards/rejected": 0.0008577358676120639,
|
| 759 |
+
"step": 480
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 1.57,
|
| 763 |
+
"learning_rate": 3.8060973297843773e-06,
|
| 764 |
+
"logits/chosen": 0.6949520111083984,
|
| 765 |
+
"logits/rejected": 0.8520463109016418,
|
| 766 |
+
"logps/chosen": -180.8011016845703,
|
| 767 |
+
"logps/rejected": -135.38951110839844,
|
| 768 |
+
"loss": 0.0001,
|
| 769 |
+
"rewards/accuracies": 0.59375,
|
| 770 |
+
"rewards/chosen": 0.0031148726120591164,
|
| 771 |
+
"rewards/margins": 0.0025527984835207462,
|
| 772 |
+
"rewards/rejected": 0.000562074186746031,
|
| 773 |
+
"step": 490
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.6,
|
| 777 |
+
"learning_rate": 3.7459599848079965e-06,
|
| 778 |
+
"logits/chosen": 0.6497796773910522,
|
| 779 |
+
"logits/rejected": 0.8080886006355286,
|
| 780 |
+
"logps/chosen": -187.10287475585938,
|
| 781 |
+
"logps/rejected": -133.8563995361328,
|
| 782 |
+
"loss": 0.0001,
|
| 783 |
+
"rewards/accuracies": 0.574999988079071,
|
| 784 |
+
"rewards/chosen": 0.004290526732802391,
|
| 785 |
+
"rewards/margins": 0.0023678760044276714,
|
| 786 |
+
"rewards/rejected": 0.0019226508447900414,
|
| 787 |
+
"step": 500
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.6,
|
| 791 |
+
"eval_logits/chosen": 0.8150748610496521,
|
| 792 |
+
"eval_logits/rejected": 0.8728834986686707,
|
| 793 |
+
"eval_logps/chosen": -256.8869323730469,
|
| 794 |
+
"eval_logps/rejected": -233.8388214111328,
|
| 795 |
+
"eval_loss": 0.0001666269963607192,
|
| 796 |
+
"eval_rewards/accuracies": 0.5170000195503235,
|
| 797 |
+
"eval_rewards/chosen": -0.0027276412583887577,
|
| 798 |
+
"eval_rewards/margins": 0.00027794469497166574,
|
| 799 |
+
"eval_rewards/rejected": -0.0030055860988795757,
|
| 800 |
+
"eval_runtime": 414.2028,
|
| 801 |
+
"eval_samples_per_second": 4.829,
|
| 802 |
+
"eval_steps_per_second": 1.207,
|
| 803 |
+
"step": 500
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 1.63,
|
| 807 |
+
"learning_rate": 3.684847614263898e-06,
|
| 808 |
+
"logits/chosen": 0.649914562702179,
|
| 809 |
+
"logits/rejected": 0.8345224261283875,
|
| 810 |
+
"logps/chosen": -170.88528442382812,
|
| 811 |
+
"logps/rejected": -113.5182876586914,
|
| 812 |
+
"loss": 0.0001,
|
| 813 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 814 |
+
"rewards/chosen": 0.004132578149437904,
|
| 815 |
+
"rewards/margins": 0.004135853610932827,
|
| 816 |
+
"rewards/rejected": -3.27564202962094e-06,
|
| 817 |
+
"step": 510
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 1.66,
|
| 821 |
+
"learning_rate": 3.622808041617133e-06,
|
| 822 |
+
"logits/chosen": 0.6696175336837769,
|
| 823 |
+
"logits/rejected": 0.8282491564750671,
|
| 824 |
+
"logps/chosen": -165.21356201171875,
|
| 825 |
+
"logps/rejected": -120.80168151855469,
|
| 826 |
+
"loss": 0.0001,
|
| 827 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 828 |
+
"rewards/chosen": 0.003804347710683942,
|
| 829 |
+
"rewards/margins": 0.002934789750725031,
|
| 830 |
+
"rewards/rejected": 0.0008695581927895546,
|
| 831 |
+
"step": 520
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 1.7,
|
| 835 |
+
"learning_rate": 3.559889815914441e-06,
|
| 836 |
+
"logits/chosen": 0.6223723888397217,
|
| 837 |
+
"logits/rejected": 0.8060038685798645,
|
| 838 |
+
"logps/chosen": -173.92562866210938,
|
| 839 |
+
"logps/rejected": -115.7896728515625,
|
| 840 |
+
"loss": 0.0001,
|
| 841 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 842 |
+
"rewards/chosen": 0.0046675438061356544,
|
| 843 |
+
"rewards/margins": 0.0034386657644063234,
|
| 844 |
+
"rewards/rejected": 0.001228878041729331,
|
| 845 |
+
"step": 530
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.73,
|
| 849 |
+
"learning_rate": 3.496142173792219e-06,
|
| 850 |
+
"logits/chosen": 0.5860522389411926,
|
| 851 |
+
"logits/rejected": 0.7538793683052063,
|
| 852 |
+
"logps/chosen": -179.20030212402344,
|
| 853 |
+
"logps/rejected": -126.5418930053711,
|
| 854 |
+
"loss": 0.0001,
|
| 855 |
+
"rewards/accuracies": 0.612500011920929,
|
| 856 |
+
"rewards/chosen": 0.0036121346056461334,
|
| 857 |
+
"rewards/margins": 0.003292589681223035,
|
| 858 |
+
"rewards/rejected": 0.00031954512814991176,
|
| 859 |
+
"step": 540
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.76,
|
| 863 |
+
"learning_rate": 3.4316150009464023e-06,
|
| 864 |
+
"logits/chosen": 0.6498968601226807,
|
| 865 |
+
"logits/rejected": 0.8035451173782349,
|
| 866 |
+
"logps/chosen": -192.52578735351562,
|
| 867 |
+
"logps/rejected": -139.58251953125,
|
| 868 |
+
"loss": 0.0001,
|
| 869 |
+
"rewards/accuracies": 0.606249988079071,
|
| 870 |
+
"rewards/chosen": 0.003925256431102753,
|
| 871 |
+
"rewards/margins": 0.0028843428008258343,
|
| 872 |
+
"rewards/rejected": 0.001040913863107562,
|
| 873 |
+
"step": 550
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 1.79,
|
| 877 |
+
"learning_rate": 3.366358793094433e-06,
|
| 878 |
+
"logits/chosen": 0.5884669423103333,
|
| 879 |
+
"logits/rejected": 0.8008158802986145,
|
| 880 |
+
"logps/chosen": -179.25753784179688,
|
| 881 |
+
"logps/rejected": -120.347412109375,
|
| 882 |
+
"loss": 0.0001,
|
| 883 |
+
"rewards/accuracies": 0.59375,
|
| 884 |
+
"rewards/chosen": 0.003893566783517599,
|
| 885 |
+
"rewards/margins": 0.0027398644015192986,
|
| 886 |
+
"rewards/rejected": 0.0011537026148289442,
|
| 887 |
+
"step": 560
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 1.82,
|
| 891 |
+
"learning_rate": 3.3004246164598535e-06,
|
| 892 |
+
"logits/chosen": 0.6389291286468506,
|
| 893 |
+
"logits/rejected": 0.8485361933708191,
|
| 894 |
+
"logps/chosen": -190.9259490966797,
|
| 895 |
+
"logps/rejected": -140.17202758789062,
|
| 896 |
+
"loss": 0.0001,
|
| 897 |
+
"rewards/accuracies": 0.675000011920929,
|
| 898 |
+
"rewards/chosen": 0.004239690490067005,
|
| 899 |
+
"rewards/margins": 0.003860587952658534,
|
| 900 |
+
"rewards/rejected": 0.0003791024792008102,
|
| 901 |
+
"step": 570
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 1.86,
|
| 905 |
+
"learning_rate": 3.233864067810446e-06,
|
| 906 |
+
"logits/chosen": 0.6482547521591187,
|
| 907 |
+
"logits/rejected": 0.8119586110115051,
|
| 908 |
+
"logps/chosen": -183.17164611816406,
|
| 909 |
+
"logps/rejected": -119.36767578125,
|
| 910 |
+
"loss": 0.0001,
|
| 911 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 912 |
+
"rewards/chosen": 0.004655472934246063,
|
| 913 |
+
"rewards/margins": 0.0037689208984375,
|
| 914 |
+
"rewards/rejected": 0.0008865518611855805,
|
| 915 |
+
"step": 580
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 1.89,
|
| 919 |
+
"learning_rate": 3.1667292340812077e-06,
|
| 920 |
+
"logits/chosen": 0.6013578176498413,
|
| 921 |
+
"logits/rejected": 0.7613108158111572,
|
| 922 |
+
"logps/chosen": -181.44009399414062,
|
| 923 |
+
"logps/rejected": -140.5045166015625,
|
| 924 |
+
"loss": 0.0001,
|
| 925 |
+
"rewards/accuracies": 0.625,
|
| 926 |
+
"rewards/chosen": 0.004445691592991352,
|
| 927 |
+
"rewards/margins": 0.003334993962198496,
|
| 928 |
+
"rewards/rejected": 0.0011106978636234999,
|
| 929 |
+
"step": 590
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 1.92,
|
| 933 |
+
"learning_rate": 3.099072651613728e-06,
|
| 934 |
+
"logits/chosen": 0.7029744386672974,
|
| 935 |
+
"logits/rejected": 0.8172439336776733,
|
| 936 |
+
"logps/chosen": -165.12625122070312,
|
| 937 |
+
"logps/rejected": -119.59954833984375,
|
| 938 |
+
"loss": 0.0001,
|
| 939 |
+
"rewards/accuracies": 0.612500011920929,
|
| 940 |
+
"rewards/chosen": 0.0032034076284617186,
|
| 941 |
+
"rewards/margins": 0.0023894021287560463,
|
| 942 |
+
"rewards/rejected": 0.0008140054414980114,
|
| 943 |
+
"step": 600
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 1.92,
|
| 947 |
+
"eval_logits/chosen": 0.817969799041748,
|
| 948 |
+
"eval_logits/rejected": 0.8756601810455322,
|
| 949 |
+
"eval_logps/chosen": -256.885986328125,
|
| 950 |
+
"eval_logps/rejected": -233.84141540527344,
|
| 951 |
+
"eval_loss": 0.00016333417443092912,
|
| 952 |
+
"eval_rewards/accuracies": 0.5070000290870667,
|
| 953 |
+
"eval_rewards/chosen": -0.0027182498015463352,
|
| 954 |
+
"eval_rewards/margins": 0.0003131589328404516,
|
| 955 |
+
"eval_rewards/rejected": -0.003031408879905939,
|
| 956 |
+
"eval_runtime": 412.1696,
|
| 957 |
+
"eval_samples_per_second": 4.852,
|
| 958 |
+
"eval_steps_per_second": 1.213,
|
| 959 |
+
"step": 600
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 1.95,
|
| 963 |
+
"learning_rate": 3.0309472650438982e-06,
|
| 964 |
+
"logits/chosen": 0.7164211869239807,
|
| 965 |
+
"logits/rejected": 0.8135250806808472,
|
| 966 |
+
"logps/chosen": -167.83938598632812,
|
| 967 |
+
"logps/rejected": -123.3685531616211,
|
| 968 |
+
"loss": 0.0001,
|
| 969 |
+
"rewards/accuracies": 0.606249988079071,
|
| 970 |
+
"rewards/chosen": 0.004080263432115316,
|
| 971 |
+
"rewards/margins": 0.002866895869374275,
|
| 972 |
+
"rewards/rejected": 0.0012133677955716848,
|
| 973 |
+
"step": 610
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 1.98,
|
| 977 |
+
"learning_rate": 2.9624063858701006e-06,
|
| 978 |
+
"logits/chosen": 0.6123205423355103,
|
| 979 |
+
"logits/rejected": 0.8067408800125122,
|
| 980 |
+
"logps/chosen": -174.879638671875,
|
| 981 |
+
"logps/rejected": -119.88896179199219,
|
| 982 |
+
"loss": 0.0001,
|
| 983 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 984 |
+
"rewards/chosen": 0.0032961233519017696,
|
| 985 |
+
"rewards/margins": 0.0031299355905503035,
|
| 986 |
+
"rewards/rejected": 0.00016618790687061846,
|
| 987 |
+
"step": 620
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 2.02,
|
| 991 |
+
"learning_rate": 2.8935036507343185e-06,
|
| 992 |
+
"logits/chosen": 0.6171947717666626,
|
| 993 |
+
"logits/rejected": 0.7731830477714539,
|
| 994 |
+
"logps/chosen": -173.52308654785156,
|
| 995 |
+
"logps/rejected": -122.30467224121094,
|
| 996 |
+
"loss": 0.0001,
|
| 997 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 998 |
+
"rewards/chosen": 0.004019217099994421,
|
| 999 |
+
"rewards/margins": 0.003546707332134247,
|
| 1000 |
+
"rewards/rejected": 0.0004725100880023092,
|
| 1001 |
+
"step": 630
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 2.05,
|
| 1005 |
+
"learning_rate": 2.8242929794487926e-06,
|
| 1006 |
+
"logits/chosen": 0.6720027923583984,
|
| 1007 |
+
"logits/rejected": 0.8209166526794434,
|
| 1008 |
+
"logps/chosen": -167.04946899414062,
|
| 1009 |
+
"logps/rejected": -120.01603698730469,
|
| 1010 |
+
"loss": 0.0001,
|
| 1011 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1012 |
+
"rewards/chosen": 0.004417883697897196,
|
| 1013 |
+
"rewards/margins": 0.00379347731359303,
|
| 1014 |
+
"rewards/rejected": 0.000624406267888844,
|
| 1015 |
+
"step": 640
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 2.08,
|
| 1019 |
+
"learning_rate": 2.7548285328010984e-06,
|
| 1020 |
+
"logits/chosen": 0.5649908781051636,
|
| 1021 |
+
"logits/rejected": 0.7522573471069336,
|
| 1022 |
+
"logps/chosen": -167.37994384765625,
|
| 1023 |
+
"logps/rejected": -118.6939697265625,
|
| 1024 |
+
"loss": 0.0001,
|
| 1025 |
+
"rewards/accuracies": 0.581250011920929,
|
| 1026 |
+
"rewards/chosen": 0.003759379032999277,
|
| 1027 |
+
"rewards/margins": 0.0030631672125309706,
|
| 1028 |
+
"rewards/rejected": 0.0006962117040529847,
|
| 1029 |
+
"step": 650
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 2.11,
|
| 1033 |
+
"learning_rate": 2.6851646701706306e-06,
|
| 1034 |
+
"logits/chosen": 0.7023177742958069,
|
| 1035 |
+
"logits/rejected": 0.8027107119560242,
|
| 1036 |
+
"logps/chosen": -177.06430053710938,
|
| 1037 |
+
"logps/rejected": -129.97714233398438,
|
| 1038 |
+
"loss": 0.0001,
|
| 1039 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1040 |
+
"rewards/chosen": 0.003546078223735094,
|
| 1041 |
+
"rewards/margins": 0.0027953279204666615,
|
| 1042 |
+
"rewards/rejected": 0.0007507502450607717,
|
| 1043 |
+
"step": 660
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 2.14,
|
| 1047 |
+
"learning_rate": 2.6153559069897007e-06,
|
| 1048 |
+
"logits/chosen": 0.6293431520462036,
|
| 1049 |
+
"logits/rejected": 0.7750530242919922,
|
| 1050 |
+
"logps/chosen": -168.621337890625,
|
| 1051 |
+
"logps/rejected": -122.5462875366211,
|
| 1052 |
+
"loss": 0.0001,
|
| 1053 |
+
"rewards/accuracies": 0.606249988079071,
|
| 1054 |
+
"rewards/chosen": 0.003033037530258298,
|
| 1055 |
+
"rewards/margins": 0.0028599021025002003,
|
| 1056 |
+
"rewards/rejected": 0.00017313548596575856,
|
| 1057 |
+
"step": 670
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 2.18,
|
| 1061 |
+
"learning_rate": 2.5454568720824937e-06,
|
| 1062 |
+
"logits/chosen": 0.5758141279220581,
|
| 1063 |
+
"logits/rejected": 0.7897475957870483,
|
| 1064 |
+
"logps/chosen": -174.54208374023438,
|
| 1065 |
+
"logps/rejected": -118.59149169921875,
|
| 1066 |
+
"loss": 0.0001,
|
| 1067 |
+
"rewards/accuracies": 0.6875,
|
| 1068 |
+
"rewards/chosen": 0.003853294998407364,
|
| 1069 |
+
"rewards/margins": 0.003840038087219,
|
| 1070 |
+
"rewards/rejected": 1.3256654710858129e-05,
|
| 1071 |
+
"step": 680
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 2.21,
|
| 1075 |
+
"learning_rate": 2.4755222649153014e-06,
|
| 1076 |
+
"logits/chosen": 0.6393508911132812,
|
| 1077 |
+
"logits/rejected": 0.8246806263923645,
|
| 1078 |
+
"logps/chosen": -184.72274780273438,
|
| 1079 |
+
"logps/rejected": -126.2255859375,
|
| 1080 |
+
"loss": 0.0001,
|
| 1081 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1082 |
+
"rewards/chosen": 0.0034170313738286495,
|
| 1083 |
+
"rewards/margins": 0.0032350593246519566,
|
| 1084 |
+
"rewards/rejected": 0.00018197241297457367,
|
| 1085 |
+
"step": 690
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 2.24,
|
| 1089 |
+
"learning_rate": 2.4056068127914803e-06,
|
| 1090 |
+
"logits/chosen": 0.7041198015213013,
|
| 1091 |
+
"logits/rejected": 0.8579221963882446,
|
| 1092 |
+
"logps/chosen": -165.37600708007812,
|
| 1093 |
+
"logps/rejected": -113.6267318725586,
|
| 1094 |
+
"loss": 0.0001,
|
| 1095 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1096 |
+
"rewards/chosen": 0.0038474693428725004,
|
| 1097 |
+
"rewards/margins": 0.004361593164503574,
|
| 1098 |
+
"rewards/rejected": -0.0005141238798387349,
|
| 1099 |
+
"step": 700
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 2.24,
|
| 1103 |
+
"eval_logits/chosen": 0.8142222166061401,
|
| 1104 |
+
"eval_logits/rejected": 0.8718712329864502,
|
| 1105 |
+
"eval_logps/chosen": -256.9122619628906,
|
| 1106 |
+
"eval_logps/rejected": -233.85919189453125,
|
| 1107 |
+
"eval_loss": 0.00016846887592691928,
|
| 1108 |
+
"eval_rewards/accuracies": 0.5065000057220459,
|
| 1109 |
+
"eval_rewards/chosen": -0.0029809277039021254,
|
| 1110 |
+
"eval_rewards/margins": 0.00022821790480520576,
|
| 1111 |
+
"eval_rewards/rejected": -0.003209145972505212,
|
| 1112 |
+
"eval_runtime": 412.3465,
|
| 1113 |
+
"eval_samples_per_second": 4.85,
|
| 1114 |
+
"eval_steps_per_second": 1.213,
|
| 1115 |
+
"step": 700
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 2.27,
|
| 1119 |
+
"learning_rate": 2.3357652280246125e-06,
|
| 1120 |
+
"logits/chosen": 0.6320663690567017,
|
| 1121 |
+
"logits/rejected": 0.785031259059906,
|
| 1122 |
+
"logps/chosen": -189.28390502929688,
|
| 1123 |
+
"logps/rejected": -144.56167602539062,
|
| 1124 |
+
"loss": 0.0001,
|
| 1125 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1126 |
+
"rewards/chosen": 0.0025972402654588223,
|
| 1127 |
+
"rewards/margins": 0.0028369042556732893,
|
| 1128 |
+
"rewards/rejected": -0.00023966425214894116,
|
| 1129 |
+
"step": 710
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 2.3,
|
| 1133 |
+
"learning_rate": 2.2660521651234036e-06,
|
| 1134 |
+
"logits/chosen": 0.6551303267478943,
|
| 1135 |
+
"logits/rejected": 0.8406831622123718,
|
| 1136 |
+
"logps/chosen": -186.629638671875,
|
| 1137 |
+
"logps/rejected": -126.31327819824219,
|
| 1138 |
+
"loss": 0.0001,
|
| 1139 |
+
"rewards/accuracies": 0.606249988079071,
|
| 1140 |
+
"rewards/chosen": 0.0038953598123043776,
|
| 1141 |
+
"rewards/margins": 0.003374651074409485,
|
| 1142 |
+
"rewards/rejected": 0.0005207090289331973,
|
| 1143 |
+
"step": 720
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 2.34,
|
| 1147 |
+
"learning_rate": 2.1965221780218173e-06,
|
| 1148 |
+
"logits/chosen": 0.6312896013259888,
|
| 1149 |
+
"logits/rejected": 0.7552987337112427,
|
| 1150 |
+
"logps/chosen": -173.728271484375,
|
| 1151 |
+
"logps/rejected": -124.27177429199219,
|
| 1152 |
+
"loss": 0.0001,
|
| 1153 |
+
"rewards/accuracies": 0.625,
|
| 1154 |
+
"rewards/chosen": 0.004026113077998161,
|
| 1155 |
+
"rewards/margins": 0.0039476132951676846,
|
| 1156 |
+
"rewards/rejected": 7.849968096707016e-05,
|
| 1157 |
+
"step": 730
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 2.37,
|
| 1161 |
+
"learning_rate": 2.1272296773879107e-06,
|
| 1162 |
+
"logits/chosen": 0.6133307814598083,
|
| 1163 |
+
"logits/rejected": 0.8152937889099121,
|
| 1164 |
+
"logps/chosen": -178.77232360839844,
|
| 1165 |
+
"logps/rejected": -125.87672424316406,
|
| 1166 |
+
"loss": 0.0001,
|
| 1167 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1168 |
+
"rewards/chosen": 0.0026045762933790684,
|
| 1169 |
+
"rewards/margins": 0.0032864962704479694,
|
| 1170 |
+
"rewards/rejected": -0.0006819200934842229,
|
| 1171 |
+
"step": 740
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 2.4,
|
| 1175 |
+
"learning_rate": 2.058228888044788e-06,
|
| 1176 |
+
"logits/chosen": 0.6092817187309265,
|
| 1177 |
+
"logits/rejected": 0.7013009190559387,
|
| 1178 |
+
"logps/chosen": -165.13404846191406,
|
| 1179 |
+
"logps/rejected": -122.30549621582031,
|
| 1180 |
+
"loss": 0.0001,
|
| 1181 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 1182 |
+
"rewards/chosen": 0.003001943463459611,
|
| 1183 |
+
"rewards/margins": 0.0025921487249433994,
|
| 1184 |
+
"rewards/rejected": 0.0004097948840353638,
|
| 1185 |
+
"step": 750
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 2.43,
|
| 1189 |
+
"learning_rate": 1.989573806536978e-06,
|
| 1190 |
+
"logits/chosen": 0.6568277478218079,
|
| 1191 |
+
"logits/rejected": 0.8574051856994629,
|
| 1192 |
+
"logps/chosen": -179.80409240722656,
|
| 1193 |
+
"logps/rejected": -120.96827697753906,
|
| 1194 |
+
"loss": 0.0001,
|
| 1195 |
+
"rewards/accuracies": 0.65625,
|
| 1196 |
+
"rewards/chosen": 0.004101686645299196,
|
| 1197 |
+
"rewards/margins": 0.00385199673473835,
|
| 1198 |
+
"rewards/rejected": 0.0002496893866918981,
|
| 1199 |
+
"step": 760
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 2.46,
|
| 1203 |
+
"learning_rate": 1.921318158875459e-06,
|
| 1204 |
+
"logits/chosen": 0.6586846113204956,
|
| 1205 |
+
"logits/rejected": 0.8662702441215515,
|
| 1206 |
+
"logps/chosen": -174.4876708984375,
|
| 1207 |
+
"logps/rejected": -128.13922119140625,
|
| 1208 |
+
"loss": 0.0001,
|
| 1209 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1210 |
+
"rewards/chosen": 0.004603301174938679,
|
| 1211 |
+
"rewards/margins": 0.0032628674525767565,
|
| 1212 |
+
"rewards/rejected": 0.0013404333731159568,
|
| 1213 |
+
"step": 770
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 2.5,
|
| 1217 |
+
"learning_rate": 1.8535153584943915e-06,
|
| 1218 |
+
"logits/chosen": 0.6371630430221558,
|
| 1219 |
+
"logits/rejected": 0.7737770080566406,
|
| 1220 |
+
"logps/chosen": -170.006591796875,
|
| 1221 |
+
"logps/rejected": -126.65934753417969,
|
| 1222 |
+
"loss": 0.0001,
|
| 1223 |
+
"rewards/accuracies": 0.59375,
|
| 1224 |
+
"rewards/chosen": 0.004735985770821571,
|
| 1225 |
+
"rewards/margins": 0.003097447333857417,
|
| 1226 |
+
"rewards/rejected": 0.001638538553379476,
|
| 1227 |
+
"step": 780
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 2.53,
|
| 1231 |
+
"learning_rate": 1.7862184644524422e-06,
|
| 1232 |
+
"logits/chosen": 0.6403388977050781,
|
| 1233 |
+
"logits/rejected": 0.7904574275016785,
|
| 1234 |
+
"logps/chosen": -178.15911865234375,
|
| 1235 |
+
"logps/rejected": -133.64309692382812,
|
| 1236 |
+
"loss": 0.0001,
|
| 1237 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1238 |
+
"rewards/chosen": 0.00445596594363451,
|
| 1239 |
+
"rewards/margins": 0.004045891109853983,
|
| 1240 |
+
"rewards/rejected": 0.0004100751248188317,
|
| 1241 |
+
"step": 790
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 2.56,
|
| 1245 |
+
"learning_rate": 1.7194801399114471e-06,
|
| 1246 |
+
"logits/chosen": 0.6325648427009583,
|
| 1247 |
+
"logits/rejected": 0.747418999671936,
|
| 1248 |
+
"logps/chosen": -174.4492950439453,
|
| 1249 |
+
"logps/rejected": -125.17801666259766,
|
| 1250 |
+
"loss": 0.0001,
|
| 1251 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 1252 |
+
"rewards/chosen": 0.004687703680247068,
|
| 1253 |
+
"rewards/margins": 0.004329483024775982,
|
| 1254 |
+
"rewards/rejected": 0.00035822103382088244,
|
| 1255 |
+
"step": 800
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 2.56,
|
| 1259 |
+
"eval_logits/chosen": 0.8135467767715454,
|
| 1260 |
+
"eval_logits/rejected": 0.8712592124938965,
|
| 1261 |
+
"eval_logps/chosen": -256.8898010253906,
|
| 1262 |
+
"eval_logps/rejected": -233.84222412109375,
|
| 1263 |
+
"eval_loss": 0.00016861619951669127,
|
| 1264 |
+
"eval_rewards/accuracies": 0.5189999938011169,
|
| 1265 |
+
"eval_rewards/chosen": -0.002756227506324649,
|
| 1266 |
+
"eval_rewards/margins": 0.00028340137214399874,
|
| 1267 |
+
"eval_rewards/rejected": -0.003039628965780139,
|
| 1268 |
+
"eval_runtime": 412.1883,
|
| 1269 |
+
"eval_samples_per_second": 4.852,
|
| 1270 |
+
"eval_steps_per_second": 1.213,
|
| 1271 |
+
"step": 800
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 2.59,
|
| 1275 |
+
"learning_rate": 1.6533526109248632e-06,
|
| 1276 |
+
"logits/chosen": 0.6175631284713745,
|
| 1277 |
+
"logits/rejected": 0.74301677942276,
|
| 1278 |
+
"logps/chosen": -179.27880859375,
|
| 1279 |
+
"logps/rejected": -123.3892822265625,
|
| 1280 |
+
"loss": 0.0001,
|
| 1281 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1282 |
+
"rewards/chosen": 0.003877174574881792,
|
| 1283 |
+
"rewards/margins": 0.003652264829725027,
|
| 1284 |
+
"rewards/rejected": 0.00022490958508569747,
|
| 1285 |
+
"step": 810
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 2.62,
|
| 1289 |
+
"learning_rate": 1.5878876255682951e-06,
|
| 1290 |
+
"logits/chosen": 0.6676111817359924,
|
| 1291 |
+
"logits/rejected": 0.7865282893180847,
|
| 1292 |
+
"logps/chosen": -189.4076385498047,
|
| 1293 |
+
"logps/rejected": -137.06271362304688,
|
| 1294 |
+
"loss": 0.0001,
|
| 1295 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1296 |
+
"rewards/chosen": 0.004411728121340275,
|
| 1297 |
+
"rewards/margins": 0.0037712506018579006,
|
| 1298 |
+
"rewards/rejected": 0.0006404774612747133,
|
| 1299 |
+
"step": 820
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 2.66,
|
| 1303 |
+
"learning_rate": 1.5231364134440485e-06,
|
| 1304 |
+
"logits/chosen": 0.6061269044876099,
|
| 1305 |
+
"logits/rejected": 0.7513775825500488,
|
| 1306 |
+
"logps/chosen": -190.88760375976562,
|
| 1307 |
+
"logps/rejected": -134.599365234375,
|
| 1308 |
+
"loss": 0.0001,
|
| 1309 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1310 |
+
"rewards/chosen": 0.004417582880705595,
|
| 1311 |
+
"rewards/margins": 0.0038115177303552628,
|
| 1312 |
+
"rewards/rejected": 0.000606064626481384,
|
| 1313 |
+
"step": 830
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 2.69,
|
| 1317 |
+
"learning_rate": 1.4591496455914292e-06,
|
| 1318 |
+
"logits/chosen": 0.6278064846992493,
|
| 1319 |
+
"logits/rejected": 0.7927815914154053,
|
| 1320 |
+
"logps/chosen": -173.77554321289062,
|
| 1321 |
+
"logps/rejected": -128.66485595703125,
|
| 1322 |
+
"loss": 0.0001,
|
| 1323 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1324 |
+
"rewards/chosen": 0.003960388712584972,
|
| 1325 |
+
"rewards/margins": 0.0037871100939810276,
|
| 1326 |
+
"rewards/rejected": 0.0001732785312924534,
|
| 1327 |
+
"step": 840
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 2.72,
|
| 1331 |
+
"learning_rate": 1.395977394834132e-06,
|
| 1332 |
+
"logits/chosen": 0.6175497174263,
|
| 1333 |
+
"logits/rejected": 0.8719260096549988,
|
| 1334 |
+
"logps/chosen": -172.75567626953125,
|
| 1335 |
+
"logps/rejected": -118.21751403808594,
|
| 1336 |
+
"loss": 0.0001,
|
| 1337 |
+
"rewards/accuracies": 0.65625,
|
| 1338 |
+
"rewards/chosen": 0.004103314597159624,
|
| 1339 |
+
"rewards/margins": 0.003662864211946726,
|
| 1340 |
+
"rewards/rejected": 0.0004404502979014069,
|
| 1341 |
+
"step": 850
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 2.75,
|
| 1345 |
+
"learning_rate": 1.3336690965957733e-06,
|
| 1346 |
+
"logits/chosen": 0.6757524013519287,
|
| 1347 |
+
"logits/rejected": 0.8124428987503052,
|
| 1348 |
+
"logps/chosen": -192.186279296875,
|
| 1349 |
+
"logps/rejected": -144.07810974121094,
|
| 1350 |
+
"loss": 0.0001,
|
| 1351 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1352 |
+
"rewards/chosen": 0.004593437071889639,
|
| 1353 |
+
"rewards/margins": 0.004751748405396938,
|
| 1354 |
+
"rewards/rejected": -0.0001583110133651644,
|
| 1355 |
+
"step": 860
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 2.78,
|
| 1359 |
+
"learning_rate": 1.2722735102142192e-06,
|
| 1360 |
+
"logits/chosen": 0.5904192328453064,
|
| 1361 |
+
"logits/rejected": 0.8412498235702515,
|
| 1362 |
+
"logps/chosen": -178.94638061523438,
|
| 1363 |
+
"logps/rejected": -116.67149353027344,
|
| 1364 |
+
"loss": 0.0001,
|
| 1365 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1366 |
+
"rewards/chosen": 0.0042373063042759895,
|
| 1367 |
+
"rewards/margins": 0.004430609289556742,
|
| 1368 |
+
"rewards/rejected": -0.00019330321811139584,
|
| 1369 |
+
"step": 870
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 2.82,
|
| 1373 |
+
"learning_rate": 1.2118386807849733e-06,
|
| 1374 |
+
"logits/chosen": 0.6753177642822266,
|
| 1375 |
+
"logits/rejected": 0.8218367695808411,
|
| 1376 |
+
"logps/chosen": -169.46713256835938,
|
| 1377 |
+
"logps/rejected": -124.23944091796875,
|
| 1378 |
+
"loss": 0.0001,
|
| 1379 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1380 |
+
"rewards/chosen": 0.0036155576817691326,
|
| 1381 |
+
"rewards/margins": 0.0030525142792612314,
|
| 1382 |
+
"rewards/rejected": 0.0005630434607155621,
|
| 1383 |
+
"step": 880
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 2.85,
|
| 1387 |
+
"learning_rate": 1.1524119015635116e-06,
|
| 1388 |
+
"logits/chosen": 0.5247241258621216,
|
| 1389 |
+
"logits/rejected": 0.7601326704025269,
|
| 1390 |
+
"logps/chosen": -187.26348876953125,
|
| 1391 |
+
"logps/rejected": -123.6650390625,
|
| 1392 |
+
"loss": 0.0001,
|
| 1393 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1394 |
+
"rewards/chosen": 0.004491317551583052,
|
| 1395 |
+
"rewards/margins": 0.003613727632910013,
|
| 1396 |
+
"rewards/rejected": 0.0008775900350883603,
|
| 1397 |
+
"step": 890
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 2.88,
|
| 1401 |
+
"learning_rate": 1.0940396769559584e-06,
|
| 1402 |
+
"logits/chosen": 0.6696980595588684,
|
| 1403 |
+
"logits/rejected": 0.7714194059371948,
|
| 1404 |
+
"logps/chosen": -191.7372283935547,
|
| 1405 |
+
"logps/rejected": -128.82192993164062,
|
| 1406 |
+
"loss": 0.0001,
|
| 1407 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1408 |
+
"rewards/chosen": 0.004666435532271862,
|
| 1409 |
+
"rewards/margins": 0.0038504847325384617,
|
| 1410 |
+
"rewards/rejected": 0.0008159511489793658,
|
| 1411 |
+
"step": 900
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 2.88,
|
| 1415 |
+
"eval_logits/chosen": 0.8135749101638794,
|
| 1416 |
+
"eval_logits/rejected": 0.8714309930801392,
|
| 1417 |
+
"eval_logps/chosen": -256.91107177734375,
|
| 1418 |
+
"eval_logps/rejected": -233.8529052734375,
|
| 1419 |
+
"eval_loss": 0.00016910216072574258,
|
| 1420 |
+
"eval_rewards/accuracies": 0.5015000104904175,
|
| 1421 |
+
"eval_rewards/chosen": -0.0029687141068279743,
|
| 1422 |
+
"eval_rewards/margins": 0.00017772088176570833,
|
| 1423 |
+
"eval_rewards/rejected": -0.0031464346684515476,
|
| 1424 |
+
"eval_runtime": 412.7326,
|
| 1425 |
+
"eval_samples_per_second": 4.846,
|
| 1426 |
+
"eval_steps_per_second": 1.211,
|
| 1427 |
+
"step": 900
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 2.91,
|
| 1431 |
+
"learning_rate": 1.036767686127079e-06,
|
| 1432 |
+
"logits/chosen": 0.6320682168006897,
|
| 1433 |
+
"logits/rejected": 0.7783384323120117,
|
| 1434 |
+
"logps/chosen": -189.1212158203125,
|
| 1435 |
+
"logps/rejected": -130.10711669921875,
|
| 1436 |
+
"loss": 0.0001,
|
| 1437 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1438 |
+
"rewards/chosen": 0.004713307600468397,
|
| 1439 |
+
"rewards/margins": 0.003414090257138014,
|
| 1440 |
+
"rewards/rejected": 0.0012992171104997396,
|
| 1441 |
+
"step": 910
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 2.94,
|
| 1445 |
+
"learning_rate": 9.806407472540644e-07,
|
| 1446 |
+
"logits/chosen": 0.6682049036026001,
|
| 1447 |
+
"logits/rejected": 0.8254071474075317,
|
| 1448 |
+
"logps/chosen": -184.09628295898438,
|
| 1449 |
+
"logps/rejected": -123.4487075805664,
|
| 1450 |
+
"loss": 0.0001,
|
| 1451 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1452 |
+
"rewards/chosen": 0.004300036001950502,
|
| 1453 |
+
"rewards/margins": 0.003941989503800869,
|
| 1454 |
+
"rewards/rejected": 0.00035804632352665067,
|
| 1455 |
+
"step": 920
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 2.98,
|
| 1459 |
+
"learning_rate": 9.257027824540823e-07,
|
| 1460 |
+
"logits/chosen": 0.6992205381393433,
|
| 1461 |
+
"logits/rejected": 0.8208533525466919,
|
| 1462 |
+
"logps/chosen": -164.10159301757812,
|
| 1463 |
+
"logps/rejected": -110.07139587402344,
|
| 1464 |
+
"loss": 0.0001,
|
| 1465 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1466 |
+
"rewards/chosen": 0.003934869542717934,
|
| 1467 |
+
"rewards/margins": 0.003884183941408992,
|
| 1468 |
+
"rewards/rejected": 5.068551399745047e-05,
|
| 1469 |
+
"step": 930
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 3.01,
|
| 1473 |
+
"learning_rate": 8.719967834130385e-07,
|
| 1474 |
+
"logits/chosen": 0.6116827726364136,
|
| 1475 |
+
"logits/rejected": 0.8160429000854492,
|
| 1476 |
+
"logps/chosen": -169.47280883789062,
|
| 1477 |
+
"logps/rejected": -115.8711166381836,
|
| 1478 |
+
"loss": 0.0001,
|
| 1479 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1480 |
+
"rewards/chosen": 0.005224294029176235,
|
| 1481 |
+
"rewards/margins": 0.004378092475235462,
|
| 1482 |
+
"rewards/rejected": 0.0008462019031867385,
|
| 1483 |
+
"step": 940
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 3.04,
|
| 1487 |
+
"learning_rate": 8.195647777424479e-07,
|
| 1488 |
+
"logits/chosen": 0.5635887384414673,
|
| 1489 |
+
"logits/rejected": 0.7940059900283813,
|
| 1490 |
+
"logps/chosen": -177.22665405273438,
|
| 1491 |
+
"logps/rejected": -120.8065414428711,
|
| 1492 |
+
"loss": 0.0001,
|
| 1493 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1494 |
+
"rewards/chosen": 0.003887483151629567,
|
| 1495 |
+
"rewards/margins": 0.0035337016452103853,
|
| 1496 |
+
"rewards/rejected": 0.00035378162283450365,
|
| 1497 |
+
"step": 950
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 3.07,
|
| 1501 |
+
"learning_rate": 7.684477960907422e-07,
|
| 1502 |
+
"logits/chosen": 0.5837413668632507,
|
| 1503 |
+
"logits/rejected": 0.7604612112045288,
|
| 1504 |
+
"logps/chosen": -175.0729217529297,
|
| 1505 |
+
"logps/rejected": -118.60954284667969,
|
| 1506 |
+
"loss": 0.0001,
|
| 1507 |
+
"rewards/accuracies": 0.625,
|
| 1508 |
+
"rewards/chosen": 0.004512741696089506,
|
| 1509 |
+
"rewards/margins": 0.0036391555331647396,
|
| 1510 |
+
"rewards/rejected": 0.0008735861629247665,
|
| 1511 |
+
"step": 960
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 3.1,
|
| 1515 |
+
"learning_rate": 7.186858400347455e-07,
|
| 1516 |
+
"logits/chosen": 0.6265005469322205,
|
| 1517 |
+
"logits/rejected": 0.7885332703590393,
|
| 1518 |
+
"logps/chosen": -182.97769165039062,
|
| 1519 |
+
"logps/rejected": -138.80723571777344,
|
| 1520 |
+
"loss": 0.0001,
|
| 1521 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 1522 |
+
"rewards/chosen": 0.004079051315784454,
|
| 1523 |
+
"rewards/margins": 0.004020996857434511,
|
| 1524 |
+
"rewards/rejected": 5.8054854889633134e-05,
|
| 1525 |
+
"step": 970
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 3.14,
|
| 1529 |
+
"learning_rate": 6.703178507764618e-07,
|
| 1530 |
+
"logits/chosen": 0.6634309887886047,
|
| 1531 |
+
"logits/rejected": 0.7914069890975952,
|
| 1532 |
+
"logps/chosen": -167.03756713867188,
|
| 1533 |
+
"logps/rejected": -124.2927474975586,
|
| 1534 |
+
"loss": 0.0001,
|
| 1535 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1536 |
+
"rewards/chosen": 0.005084961652755737,
|
| 1537 |
+
"rewards/margins": 0.0035723126493394375,
|
| 1538 |
+
"rewards/rejected": 0.0015126490034162998,
|
| 1539 |
+
"step": 980
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 3.17,
|
| 1543 |
+
"learning_rate": 6.233816786696414e-07,
|
| 1544 |
+
"logits/chosen": 0.6042094230651855,
|
| 1545 |
+
"logits/rejected": 0.7901323437690735,
|
| 1546 |
+
"logps/chosen": -195.58741760253906,
|
| 1547 |
+
"logps/rejected": -132.481689453125,
|
| 1548 |
+
"loss": 0.0001,
|
| 1549 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1550 |
+
"rewards/chosen": 0.005596310831606388,
|
| 1551 |
+
"rewards/margins": 0.0045918067917227745,
|
| 1552 |
+
"rewards/rejected": 0.001004504389129579,
|
| 1553 |
+
"step": 990
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 3.2,
|
| 1557 |
+
"learning_rate": 5.77914053600005e-07,
|
| 1558 |
+
"logits/chosen": 0.5928013920783997,
|
| 1559 |
+
"logits/rejected": 0.8330589532852173,
|
| 1560 |
+
"logps/chosen": -176.38113403320312,
|
| 1561 |
+
"logps/rejected": -123.68299865722656,
|
| 1562 |
+
"loss": 0.0001,
|
| 1563 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1564 |
+
"rewards/chosen": 0.0048389798030257225,
|
| 1565 |
+
"rewards/margins": 0.00368002662435174,
|
| 1566 |
+
"rewards/rejected": 0.001158953527919948,
|
| 1567 |
+
"step": 1000
|
| 1568 |
+
},
|
| 1569 |
+
{
|
| 1570 |
+
"epoch": 3.2,
|
| 1571 |
+
"eval_logits/chosen": 0.8156088590621948,
|
| 1572 |
+
"eval_logits/rejected": 0.8733118176460266,
|
| 1573 |
+
"eval_logps/chosen": -256.9035949707031,
|
| 1574 |
+
"eval_logps/rejected": -233.8666229248047,
|
| 1575 |
+
"eval_loss": 0.00016867661906871945,
|
| 1576 |
+
"eval_rewards/accuracies": 0.5180000066757202,
|
| 1577 |
+
"eval_rewards/chosen": -0.002894038800150156,
|
| 1578 |
+
"eval_rewards/margins": 0.0003897584683727473,
|
| 1579 |
+
"eval_rewards/rejected": -0.0032837972976267338,
|
| 1580 |
+
"eval_runtime": 412.3586,
|
| 1581 |
+
"eval_samples_per_second": 4.85,
|
| 1582 |
+
"eval_steps_per_second": 1.213,
|
| 1583 |
+
"step": 1000
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 3.23,
|
| 1587 |
+
"learning_rate": 5.339505562422851e-07,
|
| 1588 |
+
"logits/chosen": 0.6899782419204712,
|
| 1589 |
+
"logits/rejected": 0.8187691569328308,
|
| 1590 |
+
"logps/chosen": -177.58436584472656,
|
| 1591 |
+
"logps/rejected": -121.7637939453125,
|
| 1592 |
+
"loss": 0.0001,
|
| 1593 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 1594 |
+
"rewards/chosen": 0.0037464499473571777,
|
| 1595 |
+
"rewards/margins": 0.0031559974886476994,
|
| 1596 |
+
"rewards/rejected": 0.0005904529243707657,
|
| 1597 |
+
"step": 1010
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 3.26,
|
| 1601 |
+
"learning_rate": 4.915255902165734e-07,
|
| 1602 |
+
"logits/chosen": 0.6754826903343201,
|
| 1603 |
+
"logits/rejected": 0.8096501231193542,
|
| 1604 |
+
"logps/chosen": -173.49542236328125,
|
| 1605 |
+
"logps/rejected": -120.12110900878906,
|
| 1606 |
+
"loss": 0.0001,
|
| 1607 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1608 |
+
"rewards/chosen": 0.004369380883872509,
|
| 1609 |
+
"rewards/margins": 0.0036476694513112307,
|
| 1610 |
+
"rewards/rejected": 0.000721711665391922,
|
| 1611 |
+
"step": 1020
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 3.3,
|
| 1615 |
+
"learning_rate": 4.506723551657879e-07,
|
| 1616 |
+
"logits/chosen": 0.5708236694335938,
|
| 1617 |
+
"logits/rejected": 0.7320288419723511,
|
| 1618 |
+
"logps/chosen": -179.21517944335938,
|
| 1619 |
+
"logps/rejected": -131.3833465576172,
|
| 1620 |
+
"loss": 0.0001,
|
| 1621 |
+
"rewards/accuracies": 0.65625,
|
| 1622 |
+
"rewards/chosen": 0.004431615583598614,
|
| 1623 |
+
"rewards/margins": 0.003471215022727847,
|
| 1624 |
+
"rewards/rejected": 0.0009604001534171402,
|
| 1625 |
+
"step": 1030
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 3.33,
|
| 1629 |
+
"learning_rate": 4.11422820775299e-07,
|
| 1630 |
+
"logits/chosen": 0.6452796459197998,
|
| 1631 |
+
"logits/rejected": 0.8163026571273804,
|
| 1632 |
+
"logps/chosen": -205.68319702148438,
|
| 1633 |
+
"logps/rejected": -129.33828735351562,
|
| 1634 |
+
"loss": 0.0001,
|
| 1635 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1636 |
+
"rewards/chosen": 0.004141085781157017,
|
| 1637 |
+
"rewards/margins": 0.0046521080657839775,
|
| 1638 |
+
"rewards/rejected": -0.0005110226338729262,
|
| 1639 |
+
"step": 1040
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 3.36,
|
| 1643 |
+
"learning_rate": 3.7380770175506397e-07,
|
| 1644 |
+
"logits/chosen": 0.5789592862129211,
|
| 1645 |
+
"logits/rejected": 0.7904280424118042,
|
| 1646 |
+
"logps/chosen": -180.31417846679688,
|
| 1647 |
+
"logps/rejected": -129.52687072753906,
|
| 1648 |
+
"loss": 0.0001,
|
| 1649 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1650 |
+
"rewards/chosen": 0.004802372306585312,
|
| 1651 |
+
"rewards/margins": 0.004047960974276066,
|
| 1652 |
+
"rewards/rejected": 0.0007544115069322288,
|
| 1653 |
+
"step": 1050
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 3.39,
|
| 1657 |
+
"learning_rate": 3.3785643380384063e-07,
|
| 1658 |
+
"logits/chosen": 0.6884064674377441,
|
| 1659 |
+
"logits/rejected": 0.8517038226127625,
|
| 1660 |
+
"logps/chosen": -177.23248291015625,
|
| 1661 |
+
"logps/rejected": -117.1332778930664,
|
| 1662 |
+
"loss": 0.0001,
|
| 1663 |
+
"rewards/accuracies": 0.606249988079071,
|
| 1664 |
+
"rewards/chosen": 0.0034295388031750917,
|
| 1665 |
+
"rewards/margins": 0.0038840598426759243,
|
| 1666 |
+
"rewards/rejected": -0.0004545215633697808,
|
| 1667 |
+
"step": 1060
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 3.42,
|
| 1671 |
+
"learning_rate": 3.0359715057429186e-07,
|
| 1672 |
+
"logits/chosen": 0.5767031908035278,
|
| 1673 |
+
"logits/rejected": 0.6466313004493713,
|
| 1674 |
+
"logps/chosen": -172.95346069335938,
|
| 1675 |
+
"logps/rejected": -127.76350402832031,
|
| 1676 |
+
"loss": 0.0001,
|
| 1677 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1678 |
+
"rewards/chosen": 0.0043509481474757195,
|
| 1679 |
+
"rewards/margins": 0.0033387758303433657,
|
| 1680 |
+
"rewards/rejected": 0.0010121725499629974,
|
| 1681 |
+
"step": 1070
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 3.46,
|
| 1685 |
+
"learning_rate": 2.710566616570048e-07,
|
| 1686 |
+
"logits/chosen": 0.661910891532898,
|
| 1687 |
+
"logits/rejected": 0.8002559542655945,
|
| 1688 |
+
"logps/chosen": -177.2826690673828,
|
| 1689 |
+
"logps/rejected": -131.19056701660156,
|
| 1690 |
+
"loss": 0.0001,
|
| 1691 |
+
"rewards/accuracies": 0.59375,
|
| 1692 |
+
"rewards/chosen": 0.0037956517189741135,
|
| 1693 |
+
"rewards/margins": 0.003371240571141243,
|
| 1694 |
+
"rewards/rejected": 0.0004244106821715832,
|
| 1695 |
+
"step": 1080
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 3.49,
|
| 1699 |
+
"learning_rate": 2.40260431600654e-07,
|
| 1700 |
+
"logits/chosen": 0.6113812923431396,
|
| 1701 |
+
"logits/rejected": 0.8205118179321289,
|
| 1702 |
+
"logps/chosen": -174.49224853515625,
|
| 1703 |
+
"logps/rejected": -135.5587158203125,
|
| 1704 |
+
"loss": 0.0001,
|
| 1705 |
+
"rewards/accuracies": 0.574999988079071,
|
| 1706 |
+
"rewards/chosen": 0.004168152809143066,
|
| 1707 |
+
"rewards/margins": 0.0029846313409507275,
|
| 1708 |
+
"rewards/rejected": 0.0011835219338536263,
|
| 1709 |
+
"step": 1090
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 3.52,
|
| 1713 |
+
"learning_rate": 2.1123255998472952e-07,
|
| 1714 |
+
"logits/chosen": 0.6319399476051331,
|
| 1715 |
+
"logits/rejected": 0.7784051895141602,
|
| 1716 |
+
"logps/chosen": -160.6739501953125,
|
| 1717 |
+
"logps/rejected": -114.9722671508789,
|
| 1718 |
+
"loss": 0.0001,
|
| 1719 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1720 |
+
"rewards/chosen": 0.004197821486741304,
|
| 1721 |
+
"rewards/margins": 0.0031375852413475513,
|
| 1722 |
+
"rewards/rejected": 0.0010602364782243967,
|
| 1723 |
+
"step": 1100
|
| 1724 |
+
},
|
| 1725 |
+
{
|
| 1726 |
+
"epoch": 3.52,
|
| 1727 |
+
"eval_logits/chosen": 0.8145170211791992,
|
| 1728 |
+
"eval_logits/rejected": 0.8724321126937866,
|
| 1729 |
+
"eval_logps/chosen": -256.90802001953125,
|
| 1730 |
+
"eval_logps/rejected": -233.87786865234375,
|
| 1731 |
+
"eval_loss": 0.00016770198999438435,
|
| 1732 |
+
"eval_rewards/accuracies": 0.5264999866485596,
|
| 1733 |
+
"eval_rewards/chosen": -0.00293838232755661,
|
| 1734 |
+
"eval_rewards/margins": 0.00045752059668302536,
|
| 1735 |
+
"eval_rewards/rejected": -0.003395902691408992,
|
| 1736 |
+
"eval_runtime": 412.1486,
|
| 1737 |
+
"eval_samples_per_second": 4.853,
|
| 1738 |
+
"eval_steps_per_second": 1.213,
|
| 1739 |
+
"step": 1100
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 3.55,
|
| 1743 |
+
"learning_rate": 1.8399576256041525e-07,
|
| 1744 |
+
"logits/chosen": 0.6978103518486023,
|
| 1745 |
+
"logits/rejected": 0.7642599940299988,
|
| 1746 |
+
"logps/chosen": -188.5946502685547,
|
| 1747 |
+
"logps/rejected": -142.72915649414062,
|
| 1748 |
+
"loss": 0.0001,
|
| 1749 |
+
"rewards/accuracies": 0.581250011920929,
|
| 1750 |
+
"rewards/chosen": 0.004109769128262997,
|
| 1751 |
+
"rewards/margins": 0.003021980170160532,
|
| 1752 |
+
"rewards/rejected": 0.0010877888416871428,
|
| 1753 |
+
"step": 1110
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 3.58,
|
| 1757 |
+
"learning_rate": 1.58571353474391e-07,
|
| 1758 |
+
"logits/chosen": 0.6541138887405396,
|
| 1759 |
+
"logits/rejected": 0.7682735323905945,
|
| 1760 |
+
"logps/chosen": -157.97927856445312,
|
| 1761 |
+
"logps/rejected": -120.40803527832031,
|
| 1762 |
+
"loss": 0.0001,
|
| 1763 |
+
"rewards/accuracies": 0.5625,
|
| 1764 |
+
"rewards/chosen": 0.0030703709926456213,
|
| 1765 |
+
"rewards/margins": 0.0024266496766358614,
|
| 1766 |
+
"rewards/rejected": 0.0006437213160097599,
|
| 1767 |
+
"step": 1120
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 3.62,
|
| 1771 |
+
"learning_rate": 1.3497922858944857e-07,
|
| 1772 |
+
"logits/chosen": 0.7424976229667664,
|
| 1773 |
+
"logits/rejected": 0.8078197240829468,
|
| 1774 |
+
"logps/chosen": -174.2366943359375,
|
| 1775 |
+
"logps/rejected": -131.4811553955078,
|
| 1776 |
+
"loss": 0.0001,
|
| 1777 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1778 |
+
"rewards/chosen": 0.004558051936328411,
|
| 1779 |
+
"rewards/margins": 0.003959262277930975,
|
| 1780 |
+
"rewards/rejected": 0.0005987894837744534,
|
| 1781 |
+
"step": 1130
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 3.65,
|
| 1785 |
+
"learning_rate": 1.1323784991499471e-07,
|
| 1786 |
+
"logits/chosen": 0.6016920804977417,
|
| 1787 |
+
"logits/rejected": 0.7492619752883911,
|
| 1788 |
+
"logps/chosen": -172.1629638671875,
|
| 1789 |
+
"logps/rejected": -116.34178161621094,
|
| 1790 |
+
"loss": 0.0001,
|
| 1791 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 1792 |
+
"rewards/chosen": 0.004433467984199524,
|
| 1793 |
+
"rewards/margins": 0.004034010227769613,
|
| 1794 |
+
"rewards/rejected": 0.0003994574653916061,
|
| 1795 |
+
"step": 1140
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 3.68,
|
| 1799 |
+
"learning_rate": 9.336423115961002e-08,
|
| 1800 |
+
"logits/chosen": 0.6536157727241516,
|
| 1801 |
+
"logits/rejected": 0.8383282423019409,
|
| 1802 |
+
"logps/chosen": -179.27108764648438,
|
| 1803 |
+
"logps/rejected": -121.70698547363281,
|
| 1804 |
+
"loss": 0.0001,
|
| 1805 |
+
"rewards/accuracies": 0.6875,
|
| 1806 |
+
"rewards/chosen": 0.00442055519670248,
|
| 1807 |
+
"rewards/margins": 0.003978613764047623,
|
| 1808 |
+
"rewards/rejected": 0.00044194143265485764,
|
| 1809 |
+
"step": 1150
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 3.71,
|
| 1813 |
+
"learning_rate": 7.537392441697793e-08,
|
| 1814 |
+
"logits/chosen": 0.6289481520652771,
|
| 1815 |
+
"logits/rejected": 0.8159587979316711,
|
| 1816 |
+
"logps/chosen": -178.74453735351562,
|
| 1817 |
+
"logps/rejected": -126.57807922363281,
|
| 1818 |
+
"loss": 0.0001,
|
| 1819 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1820 |
+
"rewards/chosen": 0.004435301758348942,
|
| 1821 |
+
"rewards/margins": 0.0039503672160208225,
|
| 1822 |
+
"rewards/rejected": 0.0004849349206779152,
|
| 1823 |
+
"step": 1160
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 3.74,
|
| 1827 |
+
"learning_rate": 5.928100799559938e-08,
|
| 1828 |
+
"logits/chosen": 0.6689172387123108,
|
| 1829 |
+
"logits/rejected": 0.7863209843635559,
|
| 1830 |
+
"logps/chosen": -178.4945068359375,
|
| 1831 |
+
"logps/rejected": -124.7654037475586,
|
| 1832 |
+
"loss": 0.0001,
|
| 1833 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1834 |
+
"rewards/chosen": 0.005153838545084,
|
| 1835 |
+
"rewards/margins": 0.003751266747713089,
|
| 1836 |
+
"rewards/rejected": 0.001402571564540267,
|
| 1837 |
+
"step": 1170
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 3.78,
|
| 1841 |
+
"learning_rate": 4.5098075401815435e-08,
|
| 1842 |
+
"logits/chosen": 0.6192042827606201,
|
| 1843 |
+
"logits/rejected": 0.76641446352005,
|
| 1844 |
+
"logps/chosen": -171.137939453125,
|
| 1845 |
+
"logps/rejected": -125.4515151977539,
|
| 1846 |
+
"loss": 0.0001,
|
| 1847 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1848 |
+
"rewards/chosen": 0.004533619619905949,
|
| 1849 |
+
"rewards/margins": 0.003654058324173093,
|
| 1850 |
+
"rewards/rejected": 0.000879561179317534,
|
| 1851 |
+
"step": 1180
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 3.81,
|
| 1855 |
+
"learning_rate": 3.283622548476445e-08,
|
| 1856 |
+
"logits/chosen": 0.6808220148086548,
|
| 1857 |
+
"logits/rejected": 0.8677403330802917,
|
| 1858 |
+
"logps/chosen": -178.3741455078125,
|
| 1859 |
+
"logps/rejected": -125.1767349243164,
|
| 1860 |
+
"loss": 0.0001,
|
| 1861 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1862 |
+
"rewards/chosen": 0.005535735748708248,
|
| 1863 |
+
"rewards/margins": 0.004433406982570887,
|
| 1864 |
+
"rewards/rejected": 0.0011023286497220397,
|
| 1865 |
+
"step": 1190
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 3.84,
|
| 1869 |
+
"learning_rate": 2.250505375098161e-08,
|
| 1870 |
+
"logits/chosen": 0.669333815574646,
|
| 1871 |
+
"logits/rejected": 0.808275043964386,
|
| 1872 |
+
"logps/chosen": -176.30514526367188,
|
| 1873 |
+
"logps/rejected": -125.28890228271484,
|
| 1874 |
+
"loss": 0.0001,
|
| 1875 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1876 |
+
"rewards/chosen": 0.004538280423730612,
|
| 1877 |
+
"rewards/margins": 0.0036300034262239933,
|
| 1878 |
+
"rewards/rejected": 0.0009082768228836358,
|
| 1879 |
+
"step": 1200
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 3.84,
|
| 1883 |
+
"eval_logits/chosen": 0.8126665353775024,
|
| 1884 |
+
"eval_logits/rejected": 0.8705229759216309,
|
| 1885 |
+
"eval_logps/chosen": -256.92266845703125,
|
| 1886 |
+
"eval_logps/rejected": -233.87326049804688,
|
| 1887 |
+
"eval_loss": 0.00017112624482251704,
|
| 1888 |
+
"eval_rewards/accuracies": 0.5044999718666077,
|
| 1889 |
+
"eval_rewards/chosen": -0.0030848486348986626,
|
| 1890 |
+
"eval_rewards/margins": 0.0002651172399055213,
|
| 1891 |
+
"eval_rewards/rejected": -0.0033499656710773706,
|
| 1892 |
+
"eval_runtime": 412.2112,
|
| 1893 |
+
"eval_samples_per_second": 4.852,
|
| 1894 |
+
"eval_steps_per_second": 1.213,
|
| 1895 |
+
"step": 1200
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 3.87,
|
| 1899 |
+
"learning_rate": 1.4112644855438228e-08,
|
| 1900 |
+
"logits/chosen": 0.6185505390167236,
|
| 1901 |
+
"logits/rejected": 0.786270260810852,
|
| 1902 |
+
"logps/chosen": -172.150634765625,
|
| 1903 |
+
"logps/rejected": -123.99739074707031,
|
| 1904 |
+
"loss": 0.0001,
|
| 1905 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1906 |
+
"rewards/chosen": 0.005183863919228315,
|
| 1907 |
+
"rewards/margins": 0.004181054420769215,
|
| 1908 |
+
"rewards/rejected": 0.0010028090327978134,
|
| 1909 |
+
"step": 1210
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 3.9,
|
| 1913 |
+
"learning_rate": 7.665566274897007e-09,
|
| 1914 |
+
"logits/chosen": 0.6319261789321899,
|
| 1915 |
+
"logits/rejected": 0.8173401951789856,
|
| 1916 |
+
"logps/chosen": -197.80917358398438,
|
| 1917 |
+
"logps/rejected": -135.27621459960938,
|
| 1918 |
+
"loss": 0.0001,
|
| 1919 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1920 |
+
"rewards/chosen": 0.005665967706590891,
|
| 1921 |
+
"rewards/margins": 0.005035373382270336,
|
| 1922 |
+
"rewards/rejected": 0.0006305938586592674,
|
| 1923 |
+
"step": 1220
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 3.94,
|
| 1927 |
+
"learning_rate": 3.1688631685364292e-09,
|
| 1928 |
+
"logits/chosen": 0.5452470183372498,
|
| 1929 |
+
"logits/rejected": 0.7646620869636536,
|
| 1930 |
+
"logps/chosen": -184.90304565429688,
|
| 1931 |
+
"logps/rejected": -128.32806396484375,
|
| 1932 |
+
"loss": 0.0001,
|
| 1933 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1934 |
+
"rewards/chosen": 0.0048486096784472466,
|
| 1935 |
+
"rewards/margins": 0.004199582617729902,
|
| 1936 |
+
"rewards/rejected": 0.0006490271771326661,
|
| 1937 |
+
"step": 1230
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 3.97,
|
| 1941 |
+
"learning_rate": 6.260544298619664e-10,
|
| 1942 |
+
"logits/chosen": 0.6011223793029785,
|
| 1943 |
+
"logits/rejected": 0.7510023713111877,
|
| 1944 |
+
"logps/chosen": -178.15272521972656,
|
| 1945 |
+
"logps/rejected": -124.17759704589844,
|
| 1946 |
+
"loss": 0.0001,
|
| 1947 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1948 |
+
"rewards/chosen": 0.0051615191623568535,
|
| 1949 |
+
"rewards/margins": 0.003729583229869604,
|
| 1950 |
+
"rewards/rejected": 0.0014319362817332149,
|
| 1951 |
+
"step": 1240
|
| 1952 |
+
},
|
| 1953 |
+
{
|
| 1954 |
+
"epoch": 3.99,
|
| 1955 |
+
"step": 1248,
|
| 1956 |
+
"total_flos": 0.0,
|
| 1957 |
+
"train_loss": 9.35627439654733e-05,
|
| 1958 |
+
"train_runtime": 14053.8323,
|
| 1959 |
+
"train_samples_per_second": 1.423,
|
| 1960 |
+
"train_steps_per_second": 0.089
|
| 1961 |
+
}
|
| 1962 |
+
],
|
| 1963 |
+
"logging_steps": 10,
|
| 1964 |
+
"max_steps": 1248,
|
| 1965 |
+
"num_input_tokens_seen": 0,
|
| 1966 |
+
"num_train_epochs": 4,
|
| 1967 |
+
"save_steps": 100,
|
| 1968 |
+
"total_flos": 0.0,
|
| 1969 |
+
"train_batch_size": 4,
|
| 1970 |
+
"trial_name": null,
|
| 1971 |
+
"trial_params": null
|
| 1972 |
+
}
|