Model save
Browse files
README.md
ADDED
|
@@ -0,0 +1,80 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
base_model: mistralai/Mistral-7B-v0.1
|
| 9 |
+
model-index:
|
| 10 |
+
- name: zephyr-7b
|
| 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 |
+
# zephyr-7b
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on the None dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.6182
|
| 22 |
+
- Rewards/chosen: -0.4457
|
| 23 |
+
- Rewards/rejected: -0.8122
|
| 24 |
+
- Rewards/accuracies: 0.3672
|
| 25 |
+
- Rewards/margins: 0.3666
|
| 26 |
+
- Logps/rejected: -158.4149
|
| 27 |
+
- Logps/chosen: -108.4784
|
| 28 |
+
- Logits/rejected: 0.9580
|
| 29 |
+
- Logits/chosen: 0.9035
|
| 30 |
+
- Use Label: 0.0
|
| 31 |
+
- Pred Label: 0.0
|
| 32 |
+
|
| 33 |
+
## Model description
|
| 34 |
+
|
| 35 |
+
More information needed
|
| 36 |
+
|
| 37 |
+
## Intended uses & limitations
|
| 38 |
+
|
| 39 |
+
More information needed
|
| 40 |
+
|
| 41 |
+
## Training and evaluation data
|
| 42 |
+
|
| 43 |
+
More information needed
|
| 44 |
+
|
| 45 |
+
## Training procedure
|
| 46 |
+
|
| 47 |
+
### Training hyperparameters
|
| 48 |
+
|
| 49 |
+
The following hyperparameters were used during training:
|
| 50 |
+
- learning_rate: 5e-06
|
| 51 |
+
- train_batch_size: 4
|
| 52 |
+
- eval_batch_size: 8
|
| 53 |
+
- seed: 42
|
| 54 |
+
- distributed_type: multi-GPU
|
| 55 |
+
- num_devices: 8
|
| 56 |
+
- gradient_accumulation_steps: 4
|
| 57 |
+
- total_train_batch_size: 128
|
| 58 |
+
- total_eval_batch_size: 64
|
| 59 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 60 |
+
- lr_scheduler_type: cosine
|
| 61 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 62 |
+
- num_epochs: 1
|
| 63 |
+
|
| 64 |
+
### Training results
|
| 65 |
+
|
| 66 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen | Use Label | Pred Label |
|
| 67 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|:---------:|:----------:|
|
| 68 |
+
| 0.6553 | 0.21 | 100 | 0.6557 | -0.1267 | -0.2685 | 0.3633 | 0.1419 | -104.0477 | -76.5787 | -2.0726 | -2.0833 | 0.0 | 0.0 |
|
| 69 |
+
| 0.6446 | 0.42 | 200 | 0.6343 | -0.2873 | -0.5376 | 0.3828 | 0.2503 | -130.9503 | -92.6377 | -0.6864 | -0.7124 | 0.0 | 0.0 |
|
| 70 |
+
| 0.6273 | 0.63 | 300 | 0.6204 | -0.4623 | -0.7994 | 0.3672 | 0.3371 | -157.1332 | -110.1469 | 0.6726 | 0.6280 | 0.0 | 0.0 |
|
| 71 |
+
| 0.6165 | 0.84 | 400 | 0.6182 | -0.4457 | -0.8122 | 0.3672 | 0.3666 | -158.4149 | -108.4784 | 0.9580 | 0.9035 | 0.0 | 0.0 |
|
| 72 |
+
|
| 73 |
+
|
| 74 |
+
### Framework versions
|
| 75 |
+
|
| 76 |
+
- PEFT 0.7.1
|
| 77 |
+
- Transformers 4.38.2
|
| 78 |
+
- Pytorch 2.1.1+cu121
|
| 79 |
+
- Datasets 2.14.6
|
| 80 |
+
- 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 671150064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d13d352181489b0c0d490b1f94e135d6ba5d33aebeb5eb98faf9c68d8f394b98
|
| 3 |
size 671150064
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.6389844682481554,
|
| 4 |
+
"train_runtime": 9615.2592,
|
| 5 |
+
"train_samples": 61135,
|
| 6 |
+
"train_samples_per_second": 6.358,
|
| 7 |
+
"train_steps_per_second": 0.05
|
| 8 |
+
}
|
runs/Mar19_21-02-18_uclaml04.cs.ucla.edu/events.out.tfevents.1710907561.uclaml04.cs.ucla.edu.3973422.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cabb4ba084135335b7667de3d4fc284f7b992c10f172e6822a8adaca745838c7
|
| 3 |
+
size 47032
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.6389844682481554,
|
| 4 |
+
"train_runtime": 9615.2592,
|
| 5 |
+
"train_samples": 61135,
|
| 6 |
+
"train_samples_per_second": 6.358,
|
| 7 |
+
"train_steps_per_second": 0.05
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,918 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9984301412872841,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 477,
|
| 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 |
+
"grad_norm": 0.400390625,
|
| 14 |
+
"learning_rate": 1.0416666666666667e-07,
|
| 15 |
+
"logits/chosen": -2.2547454833984375,
|
| 16 |
+
"logits/rejected": -2.401865005493164,
|
| 17 |
+
"logps/chosen": -53.759212493896484,
|
| 18 |
+
"logps/rejected": -48.83185958862305,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"pred_label": 0.0,
|
| 21 |
+
"rewards/accuracies": 0.0,
|
| 22 |
+
"rewards/chosen": 0.0,
|
| 23 |
+
"rewards/margins": 0.0,
|
| 24 |
+
"rewards/rejected": 0.0,
|
| 25 |
+
"step": 1,
|
| 26 |
+
"use_label": 0.0
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.02,
|
| 30 |
+
"grad_norm": 0.4609375,
|
| 31 |
+
"learning_rate": 1.0416666666666667e-06,
|
| 32 |
+
"logits/chosen": -2.242556571960449,
|
| 33 |
+
"logits/rejected": -2.277317762374878,
|
| 34 |
+
"logps/chosen": -51.96327209472656,
|
| 35 |
+
"logps/rejected": -64.98894500732422,
|
| 36 |
+
"loss": 0.6929,
|
| 37 |
+
"pred_label": 0.0,
|
| 38 |
+
"rewards/accuracies": 0.2361111044883728,
|
| 39 |
+
"rewards/chosen": 0.002160965697839856,
|
| 40 |
+
"rewards/margins": 0.0009470728691667318,
|
| 41 |
+
"rewards/rejected": 0.0012138929450884461,
|
| 42 |
+
"step": 10,
|
| 43 |
+
"use_label": 0.0
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"epoch": 0.04,
|
| 47 |
+
"grad_norm": 0.396484375,
|
| 48 |
+
"learning_rate": 2.0833333333333334e-06,
|
| 49 |
+
"logits/chosen": -2.252474784851074,
|
| 50 |
+
"logits/rejected": -2.256141185760498,
|
| 51 |
+
"logps/chosen": -62.50165557861328,
|
| 52 |
+
"logps/rejected": -72.6328125,
|
| 53 |
+
"loss": 0.6919,
|
| 54 |
+
"pred_label": 0.0,
|
| 55 |
+
"rewards/accuracies": 0.28125,
|
| 56 |
+
"rewards/chosen": 0.01592240110039711,
|
| 57 |
+
"rewards/margins": 0.001004441175609827,
|
| 58 |
+
"rewards/rejected": 0.014917959459125996,
|
| 59 |
+
"step": 20,
|
| 60 |
+
"use_label": 0.0
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.06,
|
| 64 |
+
"grad_norm": 0.51171875,
|
| 65 |
+
"learning_rate": 3.125e-06,
|
| 66 |
+
"logits/chosen": -2.342515468597412,
|
| 67 |
+
"logits/rejected": -2.3552591800689697,
|
| 68 |
+
"logps/chosen": -79.15455627441406,
|
| 69 |
+
"logps/rejected": -98.8229751586914,
|
| 70 |
+
"loss": 0.6898,
|
| 71 |
+
"pred_label": 0.0,
|
| 72 |
+
"rewards/accuracies": 0.2874999940395355,
|
| 73 |
+
"rewards/chosen": 0.030873581767082214,
|
| 74 |
+
"rewards/margins": 0.002844910603016615,
|
| 75 |
+
"rewards/rejected": 0.02802867256104946,
|
| 76 |
+
"step": 30,
|
| 77 |
+
"use_label": 0.0
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 0.08,
|
| 81 |
+
"grad_norm": 0.51953125,
|
| 82 |
+
"learning_rate": 4.166666666666667e-06,
|
| 83 |
+
"logits/chosen": -2.323695421218872,
|
| 84 |
+
"logits/rejected": -2.3019304275512695,
|
| 85 |
+
"logps/chosen": -82.8508071899414,
|
| 86 |
+
"logps/rejected": -82.39540100097656,
|
| 87 |
+
"loss": 0.6866,
|
| 88 |
+
"pred_label": 0.0,
|
| 89 |
+
"rewards/accuracies": 0.2874999940395355,
|
| 90 |
+
"rewards/chosen": 0.033413294702768326,
|
| 91 |
+
"rewards/margins": 0.011912978254258633,
|
| 92 |
+
"rewards/rejected": 0.021500317379832268,
|
| 93 |
+
"step": 40,
|
| 94 |
+
"use_label": 0.0
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.1,
|
| 98 |
+
"grad_norm": 0.6640625,
|
| 99 |
+
"learning_rate": 4.999731868769027e-06,
|
| 100 |
+
"logits/chosen": -2.2408015727996826,
|
| 101 |
+
"logits/rejected": -2.2638282775878906,
|
| 102 |
+
"logps/chosen": -67.89698028564453,
|
| 103 |
+
"logps/rejected": -81.84117126464844,
|
| 104 |
+
"loss": 0.6805,
|
| 105 |
+
"pred_label": 0.0,
|
| 106 |
+
"rewards/accuracies": 0.32499998807907104,
|
| 107 |
+
"rewards/chosen": 0.009338948875665665,
|
| 108 |
+
"rewards/margins": 0.030354563146829605,
|
| 109 |
+
"rewards/rejected": -0.02101561427116394,
|
| 110 |
+
"step": 50,
|
| 111 |
+
"use_label": 0.0
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.13,
|
| 115 |
+
"grad_norm": 1.53125,
|
| 116 |
+
"learning_rate": 4.9903533134293035e-06,
|
| 117 |
+
"logits/chosen": -2.2194154262542725,
|
| 118 |
+
"logits/rejected": -2.1603574752807617,
|
| 119 |
+
"logps/chosen": -62.444313049316406,
|
| 120 |
+
"logps/rejected": -72.18606567382812,
|
| 121 |
+
"loss": 0.6753,
|
| 122 |
+
"pred_label": 0.0,
|
| 123 |
+
"rewards/accuracies": 0.30000001192092896,
|
| 124 |
+
"rewards/chosen": -0.027180707082152367,
|
| 125 |
+
"rewards/margins": 0.044989973306655884,
|
| 126 |
+
"rewards/rejected": -0.072170689702034,
|
| 127 |
+
"step": 60,
|
| 128 |
+
"use_label": 0.0
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.15,
|
| 132 |
+
"grad_norm": 1.84375,
|
| 133 |
+
"learning_rate": 4.967625656594782e-06,
|
| 134 |
+
"logits/chosen": -2.1111249923706055,
|
| 135 |
+
"logits/rejected": -2.109537124633789,
|
| 136 |
+
"logps/chosen": -62.041603088378906,
|
| 137 |
+
"logps/rejected": -75.64030456542969,
|
| 138 |
+
"loss": 0.666,
|
| 139 |
+
"pred_label": 0.0,
|
| 140 |
+
"rewards/accuracies": 0.25,
|
| 141 |
+
"rewards/chosen": -0.06330498307943344,
|
| 142 |
+
"rewards/margins": 0.03508424013853073,
|
| 143 |
+
"rewards/rejected": -0.09838922321796417,
|
| 144 |
+
"step": 70,
|
| 145 |
+
"use_label": 0.0
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.17,
|
| 149 |
+
"grad_norm": 1.03125,
|
| 150 |
+
"learning_rate": 4.93167072587771e-06,
|
| 151 |
+
"logits/chosen": -2.21980881690979,
|
| 152 |
+
"logits/rejected": -2.1616053581237793,
|
| 153 |
+
"logps/chosen": -60.844932556152344,
|
| 154 |
+
"logps/rejected": -74.95368957519531,
|
| 155 |
+
"loss": 0.66,
|
| 156 |
+
"pred_label": 0.0,
|
| 157 |
+
"rewards/accuracies": 0.26249998807907104,
|
| 158 |
+
"rewards/chosen": -0.12314031273126602,
|
| 159 |
+
"rewards/margins": 0.0946219339966774,
|
| 160 |
+
"rewards/rejected": -0.21776223182678223,
|
| 161 |
+
"step": 80,
|
| 162 |
+
"use_label": 0.0
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 0.19,
|
| 166 |
+
"grad_norm": 1.5390625,
|
| 167 |
+
"learning_rate": 4.882681251368549e-06,
|
| 168 |
+
"logits/chosen": -2.109405279159546,
|
| 169 |
+
"logits/rejected": -2.1181578636169434,
|
| 170 |
+
"logps/chosen": -77.24811553955078,
|
| 171 |
+
"logps/rejected": -95.32093811035156,
|
| 172 |
+
"loss": 0.6621,
|
| 173 |
+
"pred_label": 0.0,
|
| 174 |
+
"rewards/accuracies": 0.3062500059604645,
|
| 175 |
+
"rewards/chosen": -0.1886606067419052,
|
| 176 |
+
"rewards/margins": 0.07690713554620743,
|
| 177 |
+
"rewards/rejected": -0.26556771993637085,
|
| 178 |
+
"step": 90,
|
| 179 |
+
"use_label": 0.0
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.21,
|
| 183 |
+
"grad_norm": 1.1640625,
|
| 184 |
+
"learning_rate": 4.8209198325401815e-06,
|
| 185 |
+
"logits/chosen": -2.1972146034240723,
|
| 186 |
+
"logits/rejected": -2.169661283493042,
|
| 187 |
+
"logps/chosen": -92.16123962402344,
|
| 188 |
+
"logps/rejected": -84.31734466552734,
|
| 189 |
+
"loss": 0.6553,
|
| 190 |
+
"pred_label": 0.0,
|
| 191 |
+
"rewards/accuracies": 0.3499999940395355,
|
| 192 |
+
"rewards/chosen": -0.12131345272064209,
|
| 193 |
+
"rewards/margins": 0.08319222182035446,
|
| 194 |
+
"rewards/rejected": -0.20450565218925476,
|
| 195 |
+
"step": 100,
|
| 196 |
+
"use_label": 0.0
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 0.21,
|
| 200 |
+
"eval_logits/chosen": -2.0832693576812744,
|
| 201 |
+
"eval_logits/rejected": -2.0725808143615723,
|
| 202 |
+
"eval_logps/chosen": -76.57865905761719,
|
| 203 |
+
"eval_logps/rejected": -104.04773712158203,
|
| 204 |
+
"eval_loss": 0.6557236313819885,
|
| 205 |
+
"eval_pred_label": 0.0,
|
| 206 |
+
"eval_rewards/accuracies": 0.36328125,
|
| 207 |
+
"eval_rewards/chosen": -0.12666408717632294,
|
| 208 |
+
"eval_rewards/margins": 0.14188387989997864,
|
| 209 |
+
"eval_rewards/rejected": -0.26854798197746277,
|
| 210 |
+
"eval_runtime": 125.5075,
|
| 211 |
+
"eval_samples_per_second": 15.935,
|
| 212 |
+
"eval_steps_per_second": 0.255,
|
| 213 |
+
"eval_use_label": 0.0,
|
| 214 |
+
"step": 100
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 0.23,
|
| 218 |
+
"grad_norm": 1.1484375,
|
| 219 |
+
"learning_rate": 4.746717530629565e-06,
|
| 220 |
+
"logits/chosen": -2.125093460083008,
|
| 221 |
+
"logits/rejected": -2.108320713043213,
|
| 222 |
+
"logps/chosen": -86.47650146484375,
|
| 223 |
+
"logps/rejected": -108.77266693115234,
|
| 224 |
+
"loss": 0.6536,
|
| 225 |
+
"pred_label": 0.0,
|
| 226 |
+
"rewards/accuracies": 0.375,
|
| 227 |
+
"rewards/chosen": -0.1537572741508484,
|
| 228 |
+
"rewards/margins": 0.14806225895881653,
|
| 229 |
+
"rewards/rejected": -0.3018195331096649,
|
| 230 |
+
"step": 110,
|
| 231 |
+
"use_label": 0.0
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 0.25,
|
| 235 |
+
"grad_norm": 1.5390625,
|
| 236 |
+
"learning_rate": 4.660472094042121e-06,
|
| 237 |
+
"logits/chosen": -1.9497900009155273,
|
| 238 |
+
"logits/rejected": -1.8884683847427368,
|
| 239 |
+
"logps/chosen": -95.01170349121094,
|
| 240 |
+
"logps/rejected": -114.40583801269531,
|
| 241 |
+
"loss": 0.652,
|
| 242 |
+
"pred_label": 0.0,
|
| 243 |
+
"rewards/accuracies": 0.36250001192092896,
|
| 244 |
+
"rewards/chosen": -0.2506612241268158,
|
| 245 |
+
"rewards/margins": 0.16420678794384003,
|
| 246 |
+
"rewards/rejected": -0.414868026971817,
|
| 247 |
+
"step": 120,
|
| 248 |
+
"use_label": 0.0
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.27,
|
| 252 |
+
"grad_norm": 1.9296875,
|
| 253 |
+
"learning_rate": 4.5626458262912745e-06,
|
| 254 |
+
"logits/chosen": -1.7961517572402954,
|
| 255 |
+
"logits/rejected": -1.7706302404403687,
|
| 256 |
+
"logps/chosen": -90.99502563476562,
|
| 257 |
+
"logps/rejected": -112.71142578125,
|
| 258 |
+
"loss": 0.654,
|
| 259 |
+
"pred_label": 0.0,
|
| 260 |
+
"rewards/accuracies": 0.34375,
|
| 261 |
+
"rewards/chosen": -0.2521664500236511,
|
| 262 |
+
"rewards/margins": 0.1464831829071045,
|
| 263 |
+
"rewards/rejected": -0.3986496329307556,
|
| 264 |
+
"step": 130,
|
| 265 |
+
"use_label": 0.0
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.29,
|
| 269 |
+
"grad_norm": 1.9921875,
|
| 270 |
+
"learning_rate": 4.453763107901676e-06,
|
| 271 |
+
"logits/chosen": -1.7561969757080078,
|
| 272 |
+
"logits/rejected": -1.796431541442871,
|
| 273 |
+
"logps/chosen": -96.94844818115234,
|
| 274 |
+
"logps/rejected": -107.52276611328125,
|
| 275 |
+
"loss": 0.6488,
|
| 276 |
+
"pred_label": 0.0,
|
| 277 |
+
"rewards/accuracies": 0.26875001192092896,
|
| 278 |
+
"rewards/chosen": -0.1620088815689087,
|
| 279 |
+
"rewards/margins": 0.12216176092624664,
|
| 280 |
+
"rewards/rejected": -0.28417062759399414,
|
| 281 |
+
"step": 140,
|
| 282 |
+
"use_label": 0.0
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.31,
|
| 286 |
+
"grad_norm": 1.7578125,
|
| 287 |
+
"learning_rate": 4.33440758555951e-06,
|
| 288 |
+
"logits/chosen": -1.7516326904296875,
|
| 289 |
+
"logits/rejected": -1.7187411785125732,
|
| 290 |
+
"logps/chosen": -78.70259857177734,
|
| 291 |
+
"logps/rejected": -104.34063720703125,
|
| 292 |
+
"loss": 0.6451,
|
| 293 |
+
"pred_label": 0.0,
|
| 294 |
+
"rewards/accuracies": 0.32499998807907104,
|
| 295 |
+
"rewards/chosen": -0.13555890321731567,
|
| 296 |
+
"rewards/margins": 0.22945857048034668,
|
| 297 |
+
"rewards/rejected": -0.36501747369766235,
|
| 298 |
+
"step": 150,
|
| 299 |
+
"use_label": 0.0
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 0.33,
|
| 303 |
+
"grad_norm": 2.640625,
|
| 304 |
+
"learning_rate": 4.205219043576955e-06,
|
| 305 |
+
"logits/chosen": -1.481575608253479,
|
| 306 |
+
"logits/rejected": -1.468014121055603,
|
| 307 |
+
"logps/chosen": -100.68672180175781,
|
| 308 |
+
"logps/rejected": -127.04164123535156,
|
| 309 |
+
"loss": 0.6442,
|
| 310 |
+
"pred_label": 0.0,
|
| 311 |
+
"rewards/accuracies": 0.29374998807907104,
|
| 312 |
+
"rewards/chosen": -0.36356669664382935,
|
| 313 |
+
"rewards/margins": 0.1327240914106369,
|
| 314 |
+
"rewards/rejected": -0.49629077315330505,
|
| 315 |
+
"step": 160,
|
| 316 |
+
"use_label": 0.0
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.36,
|
| 320 |
+
"grad_norm": 2.390625,
|
| 321 |
+
"learning_rate": 4.066889974440757e-06,
|
| 322 |
+
"logits/chosen": -0.9005377888679504,
|
| 323 |
+
"logits/rejected": -0.8864371180534363,
|
| 324 |
+
"logps/chosen": -85.81999206542969,
|
| 325 |
+
"logps/rejected": -110.4801254272461,
|
| 326 |
+
"loss": 0.6339,
|
| 327 |
+
"pred_label": 0.0,
|
| 328 |
+
"rewards/accuracies": 0.30000001192092896,
|
| 329 |
+
"rewards/chosen": -0.3031192421913147,
|
| 330 |
+
"rewards/margins": 0.1594724804162979,
|
| 331 |
+
"rewards/rejected": -0.4625917375087738,
|
| 332 |
+
"step": 170,
|
| 333 |
+
"use_label": 0.0
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.38,
|
| 337 |
+
"grad_norm": 2.78125,
|
| 338 |
+
"learning_rate": 3.92016186682789e-06,
|
| 339 |
+
"logits/chosen": -0.591436505317688,
|
| 340 |
+
"logits/rejected": -0.5489451885223389,
|
| 341 |
+
"logps/chosen": -103.7041015625,
|
| 342 |
+
"logps/rejected": -123.32816314697266,
|
| 343 |
+
"loss": 0.6554,
|
| 344 |
+
"pred_label": 0.0,
|
| 345 |
+
"rewards/accuracies": 0.3375000059604645,
|
| 346 |
+
"rewards/chosen": -0.40916457772254944,
|
| 347 |
+
"rewards/margins": 0.2612735629081726,
|
| 348 |
+
"rewards/rejected": -0.6704381108283997,
|
| 349 |
+
"step": 180,
|
| 350 |
+
"use_label": 0.0
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.4,
|
| 354 |
+
"grad_norm": 2.09375,
|
| 355 |
+
"learning_rate": 3.7658212309857576e-06,
|
| 356 |
+
"logits/chosen": -0.801749587059021,
|
| 357 |
+
"logits/rejected": -0.588916003704071,
|
| 358 |
+
"logps/chosen": -96.86283874511719,
|
| 359 |
+
"logps/rejected": -123.17811584472656,
|
| 360 |
+
"loss": 0.6508,
|
| 361 |
+
"pred_label": 0.0,
|
| 362 |
+
"rewards/accuracies": 0.3062500059604645,
|
| 363 |
+
"rewards/chosen": -0.37751203775405884,
|
| 364 |
+
"rewards/margins": 0.21026258170604706,
|
| 365 |
+
"rewards/rejected": -0.5877746343612671,
|
| 366 |
+
"step": 190,
|
| 367 |
+
"use_label": 0.0
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.42,
|
| 371 |
+
"grad_norm": 1.59375,
|
| 372 |
+
"learning_rate": 3.604695382782159e-06,
|
| 373 |
+
"logits/chosen": -1.114527940750122,
|
| 374 |
+
"logits/rejected": -1.0130901336669922,
|
| 375 |
+
"logps/chosen": -111.54571533203125,
|
| 376 |
+
"logps/rejected": -115.97926330566406,
|
| 377 |
+
"loss": 0.6446,
|
| 378 |
+
"pred_label": 0.0,
|
| 379 |
+
"rewards/accuracies": 0.3187499940395355,
|
| 380 |
+
"rewards/chosen": -0.2986941933631897,
|
| 381 |
+
"rewards/margins": 0.1296522319316864,
|
| 382 |
+
"rewards/rejected": -0.4283464550971985,
|
| 383 |
+
"step": 200,
|
| 384 |
+
"use_label": 0.0
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.42,
|
| 388 |
+
"eval_logits/chosen": -0.7123901844024658,
|
| 389 |
+
"eval_logits/rejected": -0.6864092350006104,
|
| 390 |
+
"eval_logps/chosen": -92.6377182006836,
|
| 391 |
+
"eval_logps/rejected": -130.9503173828125,
|
| 392 |
+
"eval_loss": 0.6342783570289612,
|
| 393 |
+
"eval_pred_label": 0.0,
|
| 394 |
+
"eval_rewards/accuracies": 0.3828125,
|
| 395 |
+
"eval_rewards/chosen": -0.28725457191467285,
|
| 396 |
+
"eval_rewards/margins": 0.250319242477417,
|
| 397 |
+
"eval_rewards/rejected": -0.5375738143920898,
|
| 398 |
+
"eval_runtime": 125.6586,
|
| 399 |
+
"eval_samples_per_second": 15.916,
|
| 400 |
+
"eval_steps_per_second": 0.255,
|
| 401 |
+
"eval_use_label": 0.0,
|
| 402 |
+
"step": 200
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.44,
|
| 406 |
+
"grad_norm": 2.140625,
|
| 407 |
+
"learning_rate": 3.437648009023905e-06,
|
| 408 |
+
"logits/chosen": -0.6364002227783203,
|
| 409 |
+
"logits/rejected": -0.629191517829895,
|
| 410 |
+
"logps/chosen": -79.12034606933594,
|
| 411 |
+
"logps/rejected": -109.35395812988281,
|
| 412 |
+
"loss": 0.6319,
|
| 413 |
+
"pred_label": 0.0,
|
| 414 |
+
"rewards/accuracies": 0.35624998807907104,
|
| 415 |
+
"rewards/chosen": -0.23145589232444763,
|
| 416 |
+
"rewards/margins": 0.2322908192873001,
|
| 417 |
+
"rewards/rejected": -0.46374672651290894,
|
| 418 |
+
"step": 210,
|
| 419 |
+
"use_label": 0.0
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 0.46,
|
| 423 |
+
"grad_norm": 2.453125,
|
| 424 |
+
"learning_rate": 3.265574537815398e-06,
|
| 425 |
+
"logits/chosen": -0.24914255738258362,
|
| 426 |
+
"logits/rejected": -0.12895795702934265,
|
| 427 |
+
"logps/chosen": -123.09925842285156,
|
| 428 |
+
"logps/rejected": -127.96968078613281,
|
| 429 |
+
"loss": 0.633,
|
| 430 |
+
"pred_label": 0.0,
|
| 431 |
+
"rewards/accuracies": 0.3187499940395355,
|
| 432 |
+
"rewards/chosen": -0.43470579385757446,
|
| 433 |
+
"rewards/margins": 0.1813107430934906,
|
| 434 |
+
"rewards/rejected": -0.6160165071487427,
|
| 435 |
+
"step": 220,
|
| 436 |
+
"use_label": 0.0
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.48,
|
| 440 |
+
"grad_norm": 2.734375,
|
| 441 |
+
"learning_rate": 3.089397338773569e-06,
|
| 442 |
+
"logits/chosen": 0.08423249423503876,
|
| 443 |
+
"logits/rejected": 0.1725344955921173,
|
| 444 |
+
"logps/chosen": -98.91605377197266,
|
| 445 |
+
"logps/rejected": -125.9875259399414,
|
| 446 |
+
"loss": 0.6278,
|
| 447 |
+
"pred_label": 0.0,
|
| 448 |
+
"rewards/accuracies": 0.33125001192092896,
|
| 449 |
+
"rewards/chosen": -0.3448147773742676,
|
| 450 |
+
"rewards/margins": 0.287472665309906,
|
| 451 |
+
"rewards/rejected": -0.6322874426841736,
|
| 452 |
+
"step": 230,
|
| 453 |
+
"use_label": 0.0
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.5,
|
| 457 |
+
"grad_norm": 2.015625,
|
| 458 |
+
"learning_rate": 2.9100607788275547e-06,
|
| 459 |
+
"logits/chosen": 0.48232460021972656,
|
| 460 |
+
"logits/rejected": 0.39376580715179443,
|
| 461 |
+
"logps/chosen": -108.98759460449219,
|
| 462 |
+
"logps/rejected": -142.29344177246094,
|
| 463 |
+
"loss": 0.6294,
|
| 464 |
+
"pred_label": 0.0,
|
| 465 |
+
"rewards/accuracies": 0.3812499940395355,
|
| 466 |
+
"rewards/chosen": -0.39955058693885803,
|
| 467 |
+
"rewards/margins": 0.28114694356918335,
|
| 468 |
+
"rewards/rejected": -0.680697500705719,
|
| 469 |
+
"step": 240,
|
| 470 |
+
"use_label": 0.0
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.52,
|
| 474 |
+
"grad_norm": 2.25,
|
| 475 |
+
"learning_rate": 2.72852616010567e-06,
|
| 476 |
+
"logits/chosen": 0.35806649923324585,
|
| 477 |
+
"logits/rejected": 0.41671887040138245,
|
| 478 |
+
"logps/chosen": -126.65348052978516,
|
| 479 |
+
"logps/rejected": -151.3179168701172,
|
| 480 |
+
"loss": 0.6419,
|
| 481 |
+
"pred_label": 0.0,
|
| 482 |
+
"rewards/accuracies": 0.36250001192092896,
|
| 483 |
+
"rewards/chosen": -0.5325437784194946,
|
| 484 |
+
"rewards/margins": 0.28831106424331665,
|
| 485 |
+
"rewards/rejected": -0.8208548426628113,
|
| 486 |
+
"step": 250,
|
| 487 |
+
"use_label": 0.0
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 0.54,
|
| 491 |
+
"grad_norm": 2.46875,
|
| 492 |
+
"learning_rate": 2.5457665670441937e-06,
|
| 493 |
+
"logits/chosen": 0.4644729197025299,
|
| 494 |
+
"logits/rejected": 0.45051756501197815,
|
| 495 |
+
"logps/chosen": -110.62007904052734,
|
| 496 |
+
"logps/rejected": -142.76722717285156,
|
| 497 |
+
"loss": 0.6232,
|
| 498 |
+
"pred_label": 0.0,
|
| 499 |
+
"rewards/accuracies": 0.32499998807907104,
|
| 500 |
+
"rewards/chosen": -0.4451447129249573,
|
| 501 |
+
"rewards/margins": 0.2380482256412506,
|
| 502 |
+
"rewards/rejected": -0.6831929087638855,
|
| 503 |
+
"step": 260,
|
| 504 |
+
"use_label": 0.0
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.57,
|
| 508 |
+
"grad_norm": 2.4375,
|
| 509 |
+
"learning_rate": 2.3627616503391813e-06,
|
| 510 |
+
"logits/chosen": 0.6336380839347839,
|
| 511 |
+
"logits/rejected": 0.5556719303131104,
|
| 512 |
+
"logps/chosen": -116.7416000366211,
|
| 513 |
+
"logps/rejected": -135.33096313476562,
|
| 514 |
+
"loss": 0.6174,
|
| 515 |
+
"pred_label": 0.0,
|
| 516 |
+
"rewards/accuracies": 0.3375000059604645,
|
| 517 |
+
"rewards/chosen": -0.43825817108154297,
|
| 518 |
+
"rewards/margins": 0.22129836678504944,
|
| 519 |
+
"rewards/rejected": -0.65955650806427,
|
| 520 |
+
"step": 270,
|
| 521 |
+
"use_label": 0.0
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.59,
|
| 525 |
+
"grad_norm": 3.0625,
|
| 526 |
+
"learning_rate": 2.1804923757009885e-06,
|
| 527 |
+
"logits/chosen": 0.6383472681045532,
|
| 528 |
+
"logits/rejected": 0.7697634100914001,
|
| 529 |
+
"logps/chosen": -106.45858001708984,
|
| 530 |
+
"logps/rejected": -125.5028305053711,
|
| 531 |
+
"loss": 0.6353,
|
| 532 |
+
"pred_label": 0.0,
|
| 533 |
+
"rewards/accuracies": 0.30000001192092896,
|
| 534 |
+
"rewards/chosen": -0.4095306992530823,
|
| 535 |
+
"rewards/margins": 0.21630148589611053,
|
| 536 |
+
"rewards/rejected": -0.625832200050354,
|
| 537 |
+
"step": 280,
|
| 538 |
+
"use_label": 0.0
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.61,
|
| 542 |
+
"grad_norm": 3.328125,
|
| 543 |
+
"learning_rate": 1.9999357655598894e-06,
|
| 544 |
+
"logits/chosen": 0.1407470554113388,
|
| 545 |
+
"logits/rejected": 0.12877413630485535,
|
| 546 |
+
"logps/chosen": -108.0340805053711,
|
| 547 |
+
"logps/rejected": -136.49562072753906,
|
| 548 |
+
"loss": 0.6265,
|
| 549 |
+
"pred_label": 0.0,
|
| 550 |
+
"rewards/accuracies": 0.30000001192092896,
|
| 551 |
+
"rewards/chosen": -0.41485634446144104,
|
| 552 |
+
"rewards/margins": 0.18648667633533478,
|
| 553 |
+
"rewards/rejected": -0.601343035697937,
|
| 554 |
+
"step": 290,
|
| 555 |
+
"use_label": 0.0
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.63,
|
| 559 |
+
"grad_norm": 3.03125,
|
| 560 |
+
"learning_rate": 1.8220596619089576e-06,
|
| 561 |
+
"logits/chosen": 0.4002162516117096,
|
| 562 |
+
"logits/rejected": 0.25351682305336,
|
| 563 |
+
"logps/chosen": -127.95108795166016,
|
| 564 |
+
"logps/rejected": -172.98793029785156,
|
| 565 |
+
"loss": 0.6273,
|
| 566 |
+
"pred_label": 0.0,
|
| 567 |
+
"rewards/accuracies": 0.4124999940395355,
|
| 568 |
+
"rewards/chosen": -0.5035675168037415,
|
| 569 |
+
"rewards/margins": 0.2851078510284424,
|
| 570 |
+
"rewards/rejected": -0.7886753678321838,
|
| 571 |
+
"step": 300,
|
| 572 |
+
"use_label": 0.0
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.63,
|
| 576 |
+
"eval_logits/chosen": 0.6280341148376465,
|
| 577 |
+
"eval_logits/rejected": 0.6725929379463196,
|
| 578 |
+
"eval_logps/chosen": -110.14692687988281,
|
| 579 |
+
"eval_logps/rejected": -157.1332244873047,
|
| 580 |
+
"eval_loss": 0.620426595211029,
|
| 581 |
+
"eval_pred_label": 0.0,
|
| 582 |
+
"eval_rewards/accuracies": 0.3671875,
|
| 583 |
+
"eval_rewards/chosen": -0.46234679222106934,
|
| 584 |
+
"eval_rewards/margins": 0.33705610036849976,
|
| 585 |
+
"eval_rewards/rejected": -0.7994028329849243,
|
| 586 |
+
"eval_runtime": 125.7299,
|
| 587 |
+
"eval_samples_per_second": 15.907,
|
| 588 |
+
"eval_steps_per_second": 0.255,
|
| 589 |
+
"eval_use_label": 0.0,
|
| 590 |
+
"step": 300
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.65,
|
| 594 |
+
"grad_norm": 2.390625,
|
| 595 |
+
"learning_rate": 1.647817538357072e-06,
|
| 596 |
+
"logits/chosen": 0.33872538805007935,
|
| 597 |
+
"logits/rejected": 0.3415250778198242,
|
| 598 |
+
"logps/chosen": -95.08795166015625,
|
| 599 |
+
"logps/rejected": -142.95713806152344,
|
| 600 |
+
"loss": 0.6014,
|
| 601 |
+
"pred_label": 0.0,
|
| 602 |
+
"rewards/accuracies": 0.375,
|
| 603 |
+
"rewards/chosen": -0.39491352438926697,
|
| 604 |
+
"rewards/margins": 0.35215410590171814,
|
| 605 |
+
"rewards/rejected": -0.7470676302909851,
|
| 606 |
+
"step": 310,
|
| 607 |
+
"use_label": 0.0
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 0.67,
|
| 611 |
+
"grad_norm": 2.546875,
|
| 612 |
+
"learning_rate": 1.4781433892011132e-06,
|
| 613 |
+
"logits/chosen": 0.2642754018306732,
|
| 614 |
+
"logits/rejected": 0.4063233435153961,
|
| 615 |
+
"logps/chosen": -131.07791137695312,
|
| 616 |
+
"logps/rejected": -164.12667846679688,
|
| 617 |
+
"loss": 0.6133,
|
| 618 |
+
"pred_label": 0.0,
|
| 619 |
+
"rewards/accuracies": 0.38749998807907104,
|
| 620 |
+
"rewards/chosen": -0.580074667930603,
|
| 621 |
+
"rewards/margins": 0.38923436403274536,
|
| 622 |
+
"rewards/rejected": -0.9693089723587036,
|
| 623 |
+
"step": 320,
|
| 624 |
+
"use_label": 0.0
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.69,
|
| 628 |
+
"grad_norm": 3.15625,
|
| 629 |
+
"learning_rate": 1.3139467229135999e-06,
|
| 630 |
+
"logits/chosen": 0.5224499106407166,
|
| 631 |
+
"logits/rejected": 0.5213581919670105,
|
| 632 |
+
"logps/chosen": -130.00186157226562,
|
| 633 |
+
"logps/rejected": -156.6516876220703,
|
| 634 |
+
"loss": 0.6387,
|
| 635 |
+
"pred_label": 0.0,
|
| 636 |
+
"rewards/accuracies": 0.3375000059604645,
|
| 637 |
+
"rewards/chosen": -0.5989372134208679,
|
| 638 |
+
"rewards/margins": 0.2814994752407074,
|
| 639 |
+
"rewards/rejected": -0.8804367184638977,
|
| 640 |
+
"step": 330,
|
| 641 |
+
"use_label": 0.0
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.71,
|
| 645 |
+
"grad_norm": 2.28125,
|
| 646 |
+
"learning_rate": 1.1561076868822756e-06,
|
| 647 |
+
"logits/chosen": 0.1671726554632187,
|
| 648 |
+
"logits/rejected": 0.0974355936050415,
|
| 649 |
+
"logps/chosen": -140.3222198486328,
|
| 650 |
+
"logps/rejected": -155.46217346191406,
|
| 651 |
+
"loss": 0.6252,
|
| 652 |
+
"pred_label": 0.0,
|
| 653 |
+
"rewards/accuracies": 0.3375000059604645,
|
| 654 |
+
"rewards/chosen": -0.5558302998542786,
|
| 655 |
+
"rewards/margins": 0.23368898034095764,
|
| 656 |
+
"rewards/rejected": -0.7895193099975586,
|
| 657 |
+
"step": 340,
|
| 658 |
+
"use_label": 0.0
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 0.73,
|
| 662 |
+
"grad_norm": 3.328125,
|
| 663 |
+
"learning_rate": 1.0054723495346484e-06,
|
| 664 |
+
"logits/chosen": 0.081739641726017,
|
| 665 |
+
"logits/rejected": 0.08175826817750931,
|
| 666 |
+
"logps/chosen": -150.41506958007812,
|
| 667 |
+
"logps/rejected": -178.51565551757812,
|
| 668 |
+
"loss": 0.6231,
|
| 669 |
+
"pred_label": 0.0,
|
| 670 |
+
"rewards/accuracies": 0.375,
|
| 671 |
+
"rewards/chosen": -0.6099845170974731,
|
| 672 |
+
"rewards/margins": 0.322490930557251,
|
| 673 |
+
"rewards/rejected": -0.9324753880500793,
|
| 674 |
+
"step": 350,
|
| 675 |
+
"use_label": 0.0
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.75,
|
| 679 |
+
"grad_norm": 1.875,
|
| 680 |
+
"learning_rate": 8.628481651367876e-07,
|
| 681 |
+
"logits/chosen": 0.12279005348682404,
|
| 682 |
+
"logits/rejected": 0.20824797451496124,
|
| 683 |
+
"logps/chosen": -110.51042175292969,
|
| 684 |
+
"logps/rejected": -153.92698669433594,
|
| 685 |
+
"loss": 0.6186,
|
| 686 |
+
"pred_label": 0.0,
|
| 687 |
+
"rewards/accuracies": 0.3687500059604645,
|
| 688 |
+
"rewards/chosen": -0.46872806549072266,
|
| 689 |
+
"rewards/margins": 0.3482593894004822,
|
| 690 |
+
"rewards/rejected": -0.8169875144958496,
|
| 691 |
+
"step": 360,
|
| 692 |
+
"use_label": 0.0
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.77,
|
| 696 |
+
"grad_norm": 2.15625,
|
| 697 |
+
"learning_rate": 7.289996455765749e-07,
|
| 698 |
+
"logits/chosen": 0.19759848713874817,
|
| 699 |
+
"logits/rejected": 0.29472407698631287,
|
| 700 |
+
"logps/chosen": -103.1863021850586,
|
| 701 |
+
"logps/rejected": -143.578125,
|
| 702 |
+
"loss": 0.6166,
|
| 703 |
+
"pred_label": 0.0,
|
| 704 |
+
"rewards/accuracies": 0.3687500059604645,
|
| 705 |
+
"rewards/chosen": -0.37751519680023193,
|
| 706 |
+
"rewards/margins": 0.37911203503608704,
|
| 707 |
+
"rewards/rejected": -0.7566272020339966,
|
| 708 |
+
"step": 370,
|
| 709 |
+
"use_label": 0.0
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.8,
|
| 713 |
+
"grad_norm": 1.96875,
|
| 714 |
+
"learning_rate": 6.046442623320145e-07,
|
| 715 |
+
"logits/chosen": 0.03893072158098221,
|
| 716 |
+
"logits/rejected": 0.019468214362859726,
|
| 717 |
+
"logps/chosen": -108.17799377441406,
|
| 718 |
+
"logps/rejected": -158.08056640625,
|
| 719 |
+
"loss": 0.6183,
|
| 720 |
+
"pred_label": 0.0,
|
| 721 |
+
"rewards/accuracies": 0.3187499940395355,
|
| 722 |
+
"rewards/chosen": -0.42342591285705566,
|
| 723 |
+
"rewards/margins": 0.2937392592430115,
|
| 724 |
+
"rewards/rejected": -0.7171651124954224,
|
| 725 |
+
"step": 380,
|
| 726 |
+
"use_label": 0.0
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 0.82,
|
| 730 |
+
"grad_norm": 2.59375,
|
| 731 |
+
"learning_rate": 4.904486005914027e-07,
|
| 732 |
+
"logits/chosen": 0.33429718017578125,
|
| 733 |
+
"logits/rejected": 0.08158789575099945,
|
| 734 |
+
"logps/chosen": -151.29055786132812,
|
| 735 |
+
"logps/rejected": -180.48861694335938,
|
| 736 |
+
"loss": 0.6114,
|
| 737 |
+
"pred_label": 0.0,
|
| 738 |
+
"rewards/accuracies": 0.39375001192092896,
|
| 739 |
+
"rewards/chosen": -0.5847219824790955,
|
| 740 |
+
"rewards/margins": 0.3904651999473572,
|
| 741 |
+
"rewards/rejected": -0.9751871824264526,
|
| 742 |
+
"step": 390,
|
| 743 |
+
"use_label": 0.0
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 0.84,
|
| 747 |
+
"grad_norm": 2.015625,
|
| 748 |
+
"learning_rate": 3.8702478614051353e-07,
|
| 749 |
+
"logits/chosen": 0.126608207821846,
|
| 750 |
+
"logits/rejected": 0.2576550841331482,
|
| 751 |
+
"logps/chosen": -109.39167785644531,
|
| 752 |
+
"logps/rejected": -134.27053833007812,
|
| 753 |
+
"loss": 0.6165,
|
| 754 |
+
"pred_label": 0.0,
|
| 755 |
+
"rewards/accuracies": 0.38749998807907104,
|
| 756 |
+
"rewards/chosen": -0.36900678277015686,
|
| 757 |
+
"rewards/margins": 0.3390708863735199,
|
| 758 |
+
"rewards/rejected": -0.708077609539032,
|
| 759 |
+
"step": 400,
|
| 760 |
+
"use_label": 0.0
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 0.84,
|
| 764 |
+
"eval_logits/chosen": 0.903490424156189,
|
| 765 |
+
"eval_logits/rejected": 0.958048939704895,
|
| 766 |
+
"eval_logps/chosen": -108.47840881347656,
|
| 767 |
+
"eval_logps/rejected": -158.4149169921875,
|
| 768 |
+
"eval_loss": 0.6182093620300293,
|
| 769 |
+
"eval_pred_label": 0.0,
|
| 770 |
+
"eval_rewards/accuracies": 0.3671875,
|
| 771 |
+
"eval_rewards/chosen": -0.4456615447998047,
|
| 772 |
+
"eval_rewards/margins": 0.3665582537651062,
|
| 773 |
+
"eval_rewards/rejected": -0.8122197389602661,
|
| 774 |
+
"eval_runtime": 125.7278,
|
| 775 |
+
"eval_samples_per_second": 15.907,
|
| 776 |
+
"eval_steps_per_second": 0.255,
|
| 777 |
+
"eval_use_label": 0.0,
|
| 778 |
+
"step": 400
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.86,
|
| 782 |
+
"grad_norm": 2.046875,
|
| 783 |
+
"learning_rate": 2.9492720416985004e-07,
|
| 784 |
+
"logits/chosen": 0.39335688948631287,
|
| 785 |
+
"logits/rejected": 0.41703349351882935,
|
| 786 |
+
"logps/chosen": -106.9058837890625,
|
| 787 |
+
"logps/rejected": -138.57296752929688,
|
| 788 |
+
"loss": 0.6272,
|
| 789 |
+
"pred_label": 0.0,
|
| 790 |
+
"rewards/accuracies": 0.36250001192092896,
|
| 791 |
+
"rewards/chosen": -0.45482879877090454,
|
| 792 |
+
"rewards/margins": 0.3337084650993347,
|
| 793 |
+
"rewards/rejected": -0.788537323474884,
|
| 794 |
+
"step": 410,
|
| 795 |
+
"use_label": 0.0
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 0.88,
|
| 799 |
+
"grad_norm": 2.078125,
|
| 800 |
+
"learning_rate": 2.1464952759020857e-07,
|
| 801 |
+
"logits/chosen": 0.5264393091201782,
|
| 802 |
+
"logits/rejected": 0.4952784478664398,
|
| 803 |
+
"logps/chosen": -104.27522277832031,
|
| 804 |
+
"logps/rejected": -112.507080078125,
|
| 805 |
+
"loss": 0.6235,
|
| 806 |
+
"pred_label": 0.0,
|
| 807 |
+
"rewards/accuracies": 0.2750000059604645,
|
| 808 |
+
"rewards/chosen": -0.4333609640598297,
|
| 809 |
+
"rewards/margins": 0.1778794825077057,
|
| 810 |
+
"rewards/rejected": -0.6112405061721802,
|
| 811 |
+
"step": 420,
|
| 812 |
+
"use_label": 0.0
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.9,
|
| 816 |
+
"grad_norm": 1.734375,
|
| 817 |
+
"learning_rate": 1.4662207078575685e-07,
|
| 818 |
+
"logits/chosen": 0.47332754731178284,
|
| 819 |
+
"logits/rejected": 0.4613571763038635,
|
| 820 |
+
"logps/chosen": -144.65744018554688,
|
| 821 |
+
"logps/rejected": -170.08921813964844,
|
| 822 |
+
"loss": 0.5988,
|
| 823 |
+
"pred_label": 0.0,
|
| 824 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 825 |
+
"rewards/chosen": -0.4539059102535248,
|
| 826 |
+
"rewards/margins": 0.4534150958061218,
|
| 827 |
+
"rewards/rejected": -0.9073210954666138,
|
| 828 |
+
"step": 430,
|
| 829 |
+
"use_label": 0.0
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.92,
|
| 833 |
+
"grad_norm": 1.9609375,
|
| 834 |
+
"learning_rate": 9.120948298936422e-08,
|
| 835 |
+
"logits/chosen": 0.48202329874038696,
|
| 836 |
+
"logits/rejected": 0.6259401440620422,
|
| 837 |
+
"logps/chosen": -114.15118408203125,
|
| 838 |
+
"logps/rejected": -161.5361785888672,
|
| 839 |
+
"loss": 0.6098,
|
| 840 |
+
"pred_label": 0.0,
|
| 841 |
+
"rewards/accuracies": 0.36250001192092896,
|
| 842 |
+
"rewards/chosen": -0.4724721908569336,
|
| 843 |
+
"rewards/margins": 0.39225998520851135,
|
| 844 |
+
"rewards/rejected": -0.8647321462631226,
|
| 845 |
+
"step": 440,
|
| 846 |
+
"use_label": 0.0
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 0.94,
|
| 850 |
+
"grad_norm": 2.265625,
|
| 851 |
+
"learning_rate": 4.870879364444109e-08,
|
| 852 |
+
"logits/chosen": 0.8100695610046387,
|
| 853 |
+
"logits/rejected": 0.5903851389884949,
|
| 854 |
+
"logps/chosen": -126.81998443603516,
|
| 855 |
+
"logps/rejected": -174.6106719970703,
|
| 856 |
+
"loss": 0.6122,
|
| 857 |
+
"pred_label": 0.0,
|
| 858 |
+
"rewards/accuracies": 0.36250001192092896,
|
| 859 |
+
"rewards/chosen": -0.5456215739250183,
|
| 860 |
+
"rewards/margins": 0.3175886273384094,
|
| 861 |
+
"rewards/rejected": -0.8632103204727173,
|
| 862 |
+
"step": 450,
|
| 863 |
+
"use_label": 0.0
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.96,
|
| 867 |
+
"grad_norm": 2.140625,
|
| 868 |
+
"learning_rate": 1.93478202307823e-08,
|
| 869 |
+
"logits/chosen": 0.7001665830612183,
|
| 870 |
+
"logits/rejected": 0.7000536322593689,
|
| 871 |
+
"logps/chosen": -80.71357727050781,
|
| 872 |
+
"logps/rejected": -126.110595703125,
|
| 873 |
+
"loss": 0.6182,
|
| 874 |
+
"pred_label": 0.0,
|
| 875 |
+
"rewards/accuracies": 0.32499998807907104,
|
| 876 |
+
"rewards/chosen": -0.3459371328353882,
|
| 877 |
+
"rewards/margins": 0.2817174792289734,
|
| 878 |
+
"rewards/rejected": -0.6276546716690063,
|
| 879 |
+
"step": 460,
|
| 880 |
+
"use_label": 0.0
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"epoch": 0.98,
|
| 884 |
+
"grad_norm": 2.78125,
|
| 885 |
+
"learning_rate": 3.283947088983663e-09,
|
| 886 |
+
"logits/chosen": 0.7130995392799377,
|
| 887 |
+
"logits/rejected": 0.5145190954208374,
|
| 888 |
+
"logps/chosen": -110.40830993652344,
|
| 889 |
+
"logps/rejected": -137.49429321289062,
|
| 890 |
+
"loss": 0.6251,
|
| 891 |
+
"pred_label": 0.0,
|
| 892 |
+
"rewards/accuracies": 0.3125,
|
| 893 |
+
"rewards/chosen": -0.43079155683517456,
|
| 894 |
+
"rewards/margins": 0.25358152389526367,
|
| 895 |
+
"rewards/rejected": -0.6843730211257935,
|
| 896 |
+
"step": 470,
|
| 897 |
+
"use_label": 0.0
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.0,
|
| 901 |
+
"step": 477,
|
| 902 |
+
"total_flos": 0.0,
|
| 903 |
+
"train_loss": 0.6389844682481554,
|
| 904 |
+
"train_runtime": 9615.2592,
|
| 905 |
+
"train_samples_per_second": 6.358,
|
| 906 |
+
"train_steps_per_second": 0.05
|
| 907 |
+
}
|
| 908 |
+
],
|
| 909 |
+
"logging_steps": 10,
|
| 910 |
+
"max_steps": 477,
|
| 911 |
+
"num_input_tokens_seen": 0,
|
| 912 |
+
"num_train_epochs": 1,
|
| 913 |
+
"save_steps": 50,
|
| 914 |
+
"total_flos": 0.0,
|
| 915 |
+
"train_batch_size": 4,
|
| 916 |
+
"trial_name": null,
|
| 917 |
+
"trial_params": null
|
| 918 |
+
}
|