Model save
Browse files- README.md +81 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +891 -0
README.md
ADDED
|
@@ -0,0 +1,81 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
model-index:
|
| 9 |
+
- name: zephyr-7b-dpo-full-ultrabin-reward-scale-1-rpo
|
| 10 |
+
results: []
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 14 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 15 |
+
|
| 16 |
+
# zephyr-7b-dpo-full-ultrabin-reward-scale-1-rpo
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on an unknown dataset.
|
| 19 |
+
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.0345
|
| 21 |
+
- Rewards/chosen: -0.1240
|
| 22 |
+
- Rewards/rejected: -0.4158
|
| 23 |
+
- Rewards/accuracies: 0.7734
|
| 24 |
+
- Rewards/margins: 0.2918
|
| 25 |
+
- Logps/rejected: -304.2409
|
| 26 |
+
- Logps/chosen: -275.0308
|
| 27 |
+
- Logits/rejected: -2.4195
|
| 28 |
+
- Logits/chosen: -2.5003
|
| 29 |
+
|
| 30 |
+
## Model description
|
| 31 |
+
|
| 32 |
+
More information needed
|
| 33 |
+
|
| 34 |
+
## Intended uses & limitations
|
| 35 |
+
|
| 36 |
+
More information needed
|
| 37 |
+
|
| 38 |
+
## Training and evaluation data
|
| 39 |
+
|
| 40 |
+
More information needed
|
| 41 |
+
|
| 42 |
+
## Training procedure
|
| 43 |
+
|
| 44 |
+
### Training hyperparameters
|
| 45 |
+
|
| 46 |
+
The following hyperparameters were used during training:
|
| 47 |
+
- learning_rate: 5e-07
|
| 48 |
+
- train_batch_size: 8
|
| 49 |
+
- eval_batch_size: 8
|
| 50 |
+
- seed: 55
|
| 51 |
+
- distributed_type: multi-GPU
|
| 52 |
+
- num_devices: 8
|
| 53 |
+
- gradient_accumulation_steps: 2
|
| 54 |
+
- total_train_batch_size: 128
|
| 55 |
+
- total_eval_batch_size: 64
|
| 56 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 57 |
+
- lr_scheduler_type: cosine
|
| 58 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 59 |
+
- num_epochs: 1
|
| 60 |
+
|
| 61 |
+
### Training results
|
| 62 |
+
|
| 63 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 64 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 65 |
+
| 0.0497 | 0.1046 | 50 | 0.0471 | 0.0341 | -0.0639 | 0.6953 | 0.0980 | -269.0505 | -259.2180 | -2.5747 | -2.6118 |
|
| 66 |
+
| 0.0399 | 0.2092 | 100 | 0.0400 | -0.0674 | -0.3039 | 0.7656 | 0.2365 | -293.0492 | -269.3653 | -2.2263 | -2.2796 |
|
| 67 |
+
| 0.0384 | 0.3138 | 150 | 0.0368 | -0.1521 | -0.4051 | 0.7812 | 0.2530 | -303.1761 | -277.8396 | -2.4575 | -2.5017 |
|
| 68 |
+
| 0.0354 | 0.4184 | 200 | 0.0368 | -0.1608 | -0.4413 | 0.7812 | 0.2805 | -306.7949 | -278.7134 | -2.6355 | -2.6785 |
|
| 69 |
+
| 0.035 | 0.5230 | 250 | 0.0359 | -0.0276 | -0.3002 | 0.7812 | 0.2726 | -292.6817 | -265.3905 | -2.5364 | -2.5931 |
|
| 70 |
+
| 0.0336 | 0.6276 | 300 | 0.0351 | -0.1609 | -0.4489 | 0.7734 | 0.2880 | -307.5566 | -278.7195 | -2.3179 | -2.4060 |
|
| 71 |
+
| 0.0338 | 0.7322 | 350 | 0.0348 | -0.1145 | -0.3940 | 0.7695 | 0.2795 | -302.0604 | -274.0787 | -2.3603 | -2.4329 |
|
| 72 |
+
| 0.0352 | 0.8368 | 400 | 0.0345 | -0.1250 | -0.4112 | 0.7734 | 0.2863 | -303.7862 | -275.1277 | -2.4372 | -2.5111 |
|
| 73 |
+
| 0.0342 | 0.9414 | 450 | 0.0345 | -0.1240 | -0.4158 | 0.7734 | 0.2918 | -304.2409 | -275.0308 | -2.4195 | -2.5003 |
|
| 74 |
+
|
| 75 |
+
|
| 76 |
+
### Framework versions
|
| 77 |
+
|
| 78 |
+
- Transformers 4.44.0.dev0
|
| 79 |
+
- Pytorch 2.1.2
|
| 80 |
+
- Datasets 2.20.0
|
| 81 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.037969033089641745,
|
| 5 |
+
"train_runtime": 12745.9722,
|
| 6 |
+
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 4.796,
|
| 8 |
+
"train_steps_per_second": 0.038
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"transformers_version": "4.44.0.dev0"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.037969033089641745,
|
| 5 |
+
"train_runtime": 12745.9722,
|
| 6 |
+
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 4.796,
|
| 8 |
+
"train_steps_per_second": 0.038
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,891 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 50,
|
| 6 |
+
"global_step": 478,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.02092050209205021,
|
| 13 |
+
"grad_norm": 3.4774240176403444,
|
| 14 |
+
"learning_rate": 1.0416666666666667e-07,
|
| 15 |
+
"logits/chosen": -2.6303696632385254,
|
| 16 |
+
"logits/rejected": -2.576477289199829,
|
| 17 |
+
"logps/chosen": -288.6245422363281,
|
| 18 |
+
"logps/rejected": -275.9084167480469,
|
| 19 |
+
"loss": 0.0582,
|
| 20 |
+
"rewards/accuracies": 0.46875,
|
| 21 |
+
"rewards/chosen": 0.0006722577381879091,
|
| 22 |
+
"rewards/margins": 0.001247679116204381,
|
| 23 |
+
"rewards/rejected": -0.0005754214362241328,
|
| 24 |
+
"step": 10
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.04184100418410042,
|
| 28 |
+
"grad_norm": 3.1260461200672642,
|
| 29 |
+
"learning_rate": 2.0833333333333333e-07,
|
| 30 |
+
"logits/chosen": -2.6448445320129395,
|
| 31 |
+
"logits/rejected": -2.6134140491485596,
|
| 32 |
+
"logps/chosen": -293.53070068359375,
|
| 33 |
+
"logps/rejected": -259.20135498046875,
|
| 34 |
+
"loss": 0.0585,
|
| 35 |
+
"rewards/accuracies": 0.550000011920929,
|
| 36 |
+
"rewards/chosen": 0.0029079406522214413,
|
| 37 |
+
"rewards/margins": 0.001445380854420364,
|
| 38 |
+
"rewards/rejected": 0.0014625597978010774,
|
| 39 |
+
"step": 20
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.06276150627615062,
|
| 43 |
+
"grad_norm": 2.8396874218169663,
|
| 44 |
+
"learning_rate": 3.1249999999999997e-07,
|
| 45 |
+
"logits/chosen": -2.666360855102539,
|
| 46 |
+
"logits/rejected": -2.589587688446045,
|
| 47 |
+
"logps/chosen": -294.760986328125,
|
| 48 |
+
"logps/rejected": -287.2416076660156,
|
| 49 |
+
"loss": 0.0569,
|
| 50 |
+
"rewards/accuracies": 0.59375,
|
| 51 |
+
"rewards/chosen": 0.01464819349348545,
|
| 52 |
+
"rewards/margins": 0.007496826350688934,
|
| 53 |
+
"rewards/rejected": 0.007151367608457804,
|
| 54 |
+
"step": 30
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.08368200836820083,
|
| 58 |
+
"grad_norm": 2.618292822796603,
|
| 59 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 60 |
+
"logits/chosen": -2.6354165077209473,
|
| 61 |
+
"logits/rejected": -2.552605152130127,
|
| 62 |
+
"logps/chosen": -270.3968200683594,
|
| 63 |
+
"logps/rejected": -240.0804443359375,
|
| 64 |
+
"loss": 0.0544,
|
| 65 |
+
"rewards/accuracies": 0.625,
|
| 66 |
+
"rewards/chosen": 0.03499267250299454,
|
| 67 |
+
"rewards/margins": 0.025343021377921104,
|
| 68 |
+
"rewards/rejected": 0.009649652987718582,
|
| 69 |
+
"step": 40
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.10460251046025104,
|
| 73 |
+
"grad_norm": 3.012648304181472,
|
| 74 |
+
"learning_rate": 4.999733114418725e-07,
|
| 75 |
+
"logits/chosen": -2.5784354209899902,
|
| 76 |
+
"logits/rejected": -2.5706632137298584,
|
| 77 |
+
"logps/chosen": -262.97918701171875,
|
| 78 |
+
"logps/rejected": -244.66690063476562,
|
| 79 |
+
"loss": 0.0497,
|
| 80 |
+
"rewards/accuracies": 0.699999988079071,
|
| 81 |
+
"rewards/chosen": 0.0244884192943573,
|
| 82 |
+
"rewards/margins": 0.06963478773832321,
|
| 83 |
+
"rewards/rejected": -0.04514636844396591,
|
| 84 |
+
"step": 50
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.10460251046025104,
|
| 88 |
+
"eval_logits/chosen": -2.6117944717407227,
|
| 89 |
+
"eval_logits/rejected": -2.5747056007385254,
|
| 90 |
+
"eval_logps/chosen": -259.21795654296875,
|
| 91 |
+
"eval_logps/rejected": -269.0505065917969,
|
| 92 |
+
"eval_loss": 0.047118376940488815,
|
| 93 |
+
"eval_rewards/accuracies": 0.6953125,
|
| 94 |
+
"eval_rewards/chosen": 0.034119848161935806,
|
| 95 |
+
"eval_rewards/margins": 0.09800390154123306,
|
| 96 |
+
"eval_rewards/rejected": -0.06388404965400696,
|
| 97 |
+
"eval_runtime": 104.2768,
|
| 98 |
+
"eval_samples_per_second": 19.18,
|
| 99 |
+
"eval_steps_per_second": 0.307,
|
| 100 |
+
"step": 50
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.12552301255230125,
|
| 104 |
+
"grad_norm": 3.174321476078958,
|
| 105 |
+
"learning_rate": 4.990398100856366e-07,
|
| 106 |
+
"logits/chosen": -2.537804126739502,
|
| 107 |
+
"logits/rejected": -2.49711012840271,
|
| 108 |
+
"logps/chosen": -261.7084045410156,
|
| 109 |
+
"logps/rejected": -251.53125,
|
| 110 |
+
"loss": 0.0464,
|
| 111 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 112 |
+
"rewards/chosen": 0.0611313059926033,
|
| 113 |
+
"rewards/margins": 0.11455783993005753,
|
| 114 |
+
"rewards/rejected": -0.053426533937454224,
|
| 115 |
+
"step": 60
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.14644351464435146,
|
| 119 |
+
"grad_norm": 3.5406121572748104,
|
| 120 |
+
"learning_rate": 4.967775735898179e-07,
|
| 121 |
+
"logits/chosen": -2.5840494632720947,
|
| 122 |
+
"logits/rejected": -2.512150764465332,
|
| 123 |
+
"logps/chosen": -288.1974792480469,
|
| 124 |
+
"logps/rejected": -287.8002014160156,
|
| 125 |
+
"loss": 0.0445,
|
| 126 |
+
"rewards/accuracies": 0.675000011920929,
|
| 127 |
+
"rewards/chosen": -0.03117268718779087,
|
| 128 |
+
"rewards/margins": 0.14650143682956696,
|
| 129 |
+
"rewards/rejected": -0.17767412960529327,
|
| 130 |
+
"step": 70
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.16736401673640167,
|
| 134 |
+
"grad_norm": 3.2391048621845417,
|
| 135 |
+
"learning_rate": 4.931986719649298e-07,
|
| 136 |
+
"logits/chosen": -2.4763026237487793,
|
| 137 |
+
"logits/rejected": -2.464881658554077,
|
| 138 |
+
"logps/chosen": -260.2668762207031,
|
| 139 |
+
"logps/rejected": -276.872314453125,
|
| 140 |
+
"loss": 0.0414,
|
| 141 |
+
"rewards/accuracies": 0.637499988079071,
|
| 142 |
+
"rewards/chosen": -0.011183048598468304,
|
| 143 |
+
"rewards/margins": 0.13343258202075958,
|
| 144 |
+
"rewards/rejected": -0.14461562037467957,
|
| 145 |
+
"step": 80
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.18828451882845187,
|
| 149 |
+
"grad_norm": 3.794320711197701,
|
| 150 |
+
"learning_rate": 4.883222001996351e-07,
|
| 151 |
+
"logits/chosen": -2.135620355606079,
|
| 152 |
+
"logits/rejected": -2.0978758335113525,
|
| 153 |
+
"logps/chosen": -281.79473876953125,
|
| 154 |
+
"logps/rejected": -288.814697265625,
|
| 155 |
+
"loss": 0.041,
|
| 156 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 157 |
+
"rewards/chosen": -0.10697118937969208,
|
| 158 |
+
"rewards/margins": 0.17307965457439423,
|
| 159 |
+
"rewards/rejected": -0.2800508439540863,
|
| 160 |
+
"step": 90
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.20920502092050208,
|
| 164 |
+
"grad_norm": 4.607538199967544,
|
| 165 |
+
"learning_rate": 4.821741763807186e-07,
|
| 166 |
+
"logits/chosen": -2.0615501403808594,
|
| 167 |
+
"logits/rejected": -1.8890082836151123,
|
| 168 |
+
"logps/chosen": -300.027587890625,
|
| 169 |
+
"logps/rejected": -253.22848510742188,
|
| 170 |
+
"loss": 0.0399,
|
| 171 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 172 |
+
"rewards/chosen": -0.06046273559331894,
|
| 173 |
+
"rewards/margins": 0.20909741520881653,
|
| 174 |
+
"rewards/rejected": -0.26956015825271606,
|
| 175 |
+
"step": 100
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.20920502092050208,
|
| 179 |
+
"eval_logits/chosen": -2.279555320739746,
|
| 180 |
+
"eval_logits/rejected": -2.226322650909424,
|
| 181 |
+
"eval_logps/chosen": -269.3652648925781,
|
| 182 |
+
"eval_logps/rejected": -293.0491638183594,
|
| 183 |
+
"eval_loss": 0.03997171297669411,
|
| 184 |
+
"eval_rewards/accuracies": 0.765625,
|
| 185 |
+
"eval_rewards/chosen": -0.06735337525606155,
|
| 186 |
+
"eval_rewards/margins": 0.23651722073554993,
|
| 187 |
+
"eval_rewards/rejected": -0.30387061834335327,
|
| 188 |
+
"eval_runtime": 102.9716,
|
| 189 |
+
"eval_samples_per_second": 19.423,
|
| 190 |
+
"eval_steps_per_second": 0.311,
|
| 191 |
+
"step": 100
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.2301255230125523,
|
| 195 |
+
"grad_norm": 3.018182648285734,
|
| 196 |
+
"learning_rate": 4.747874028753375e-07,
|
| 197 |
+
"logits/chosen": -2.2557241916656494,
|
| 198 |
+
"logits/rejected": -2.202664375305176,
|
| 199 |
+
"logps/chosen": -309.6082458496094,
|
| 200 |
+
"logps/rejected": -311.09197998046875,
|
| 201 |
+
"loss": 0.0395,
|
| 202 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 203 |
+
"rewards/chosen": -0.14871956408023834,
|
| 204 |
+
"rewards/margins": 0.2059335708618164,
|
| 205 |
+
"rewards/rejected": -0.35465317964553833,
|
| 206 |
+
"step": 110
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.2510460251046025,
|
| 210 |
+
"grad_norm": 3.013542760925059,
|
| 211 |
+
"learning_rate": 4.662012913161997e-07,
|
| 212 |
+
"logits/chosen": -2.4675145149230957,
|
| 213 |
+
"logits/rejected": -2.377410411834717,
|
| 214 |
+
"logps/chosen": -294.02093505859375,
|
| 215 |
+
"logps/rejected": -280.5357971191406,
|
| 216 |
+
"loss": 0.0383,
|
| 217 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 218 |
+
"rewards/chosen": -0.17444203794002533,
|
| 219 |
+
"rewards/margins": 0.2464580088853836,
|
| 220 |
+
"rewards/rejected": -0.42090004682540894,
|
| 221 |
+
"step": 120
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.2719665271966527,
|
| 225 |
+
"grad_norm": 3.0191370619156745,
|
| 226 |
+
"learning_rate": 4.5646165232345103e-07,
|
| 227 |
+
"logits/chosen": -2.4819345474243164,
|
| 228 |
+
"logits/rejected": -2.4598195552825928,
|
| 229 |
+
"logps/chosen": -303.03009033203125,
|
| 230 |
+
"logps/rejected": -293.29705810546875,
|
| 231 |
+
"loss": 0.037,
|
| 232 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 233 |
+
"rewards/chosen": -0.05563073605298996,
|
| 234 |
+
"rewards/margins": 0.2324099987745285,
|
| 235 |
+
"rewards/rejected": -0.28804072737693787,
|
| 236 |
+
"step": 130
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.2928870292887029,
|
| 240 |
+
"grad_norm": 3.626977578421201,
|
| 241 |
+
"learning_rate": 4.456204510851956e-07,
|
| 242 |
+
"logits/chosen": -2.478722095489502,
|
| 243 |
+
"logits/rejected": -2.3737692832946777,
|
| 244 |
+
"logps/chosen": -311.2557067871094,
|
| 245 |
+
"logps/rejected": -308.92388916015625,
|
| 246 |
+
"loss": 0.0365,
|
| 247 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 248 |
+
"rewards/chosen": -0.10953263938426971,
|
| 249 |
+
"rewards/margins": 0.28694862127304077,
|
| 250 |
+
"rewards/rejected": -0.3964812457561493,
|
| 251 |
+
"step": 140
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.3138075313807531,
|
| 255 |
+
"grad_norm": 4.252526149590865,
|
| 256 |
+
"learning_rate": 4.337355301007335e-07,
|
| 257 |
+
"logits/chosen": -2.488487482070923,
|
| 258 |
+
"logits/rejected": -2.339409351348877,
|
| 259 |
+
"logps/chosen": -307.0581970214844,
|
| 260 |
+
"logps/rejected": -306.7461853027344,
|
| 261 |
+
"loss": 0.0384,
|
| 262 |
+
"rewards/accuracies": 0.731249988079071,
|
| 263 |
+
"rewards/chosen": -0.16012495756149292,
|
| 264 |
+
"rewards/margins": 0.21382689476013184,
|
| 265 |
+
"rewards/rejected": -0.37395185232162476,
|
| 266 |
+
"step": 150
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.3138075313807531,
|
| 270 |
+
"eval_logits/chosen": -2.501673460006714,
|
| 271 |
+
"eval_logits/rejected": -2.4575157165527344,
|
| 272 |
+
"eval_logps/chosen": -277.8395690917969,
|
| 273 |
+
"eval_logps/rejected": -303.1760559082031,
|
| 274 |
+
"eval_loss": 0.03684209659695625,
|
| 275 |
+
"eval_rewards/accuracies": 0.78125,
|
| 276 |
+
"eval_rewards/chosen": -0.1520964801311493,
|
| 277 |
+
"eval_rewards/margins": 0.2530430555343628,
|
| 278 |
+
"eval_rewards/rejected": -0.4051395058631897,
|
| 279 |
+
"eval_runtime": 102.7099,
|
| 280 |
+
"eval_samples_per_second": 19.472,
|
| 281 |
+
"eval_steps_per_second": 0.312,
|
| 282 |
+
"step": 150
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.33472803347280333,
|
| 286 |
+
"grad_norm": 3.258158347527928,
|
| 287 |
+
"learning_rate": 4.2087030056579986e-07,
|
| 288 |
+
"logits/chosen": -2.4619104862213135,
|
| 289 |
+
"logits/rejected": -2.3939900398254395,
|
| 290 |
+
"logps/chosen": -278.5140686035156,
|
| 291 |
+
"logps/rejected": -272.5802307128906,
|
| 292 |
+
"loss": 0.0367,
|
| 293 |
+
"rewards/accuracies": 0.75,
|
| 294 |
+
"rewards/chosen": -0.07991068065166473,
|
| 295 |
+
"rewards/margins": 0.2492825984954834,
|
| 296 |
+
"rewards/rejected": -0.3291932940483093,
|
| 297 |
+
"step": 160
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.35564853556485354,
|
| 301 |
+
"grad_norm": 5.659535165307617,
|
| 302 |
+
"learning_rate": 4.070934040463998e-07,
|
| 303 |
+
"logits/chosen": -2.3733975887298584,
|
| 304 |
+
"logits/rejected": -2.274611711502075,
|
| 305 |
+
"logps/chosen": -280.55621337890625,
|
| 306 |
+
"logps/rejected": -291.4908142089844,
|
| 307 |
+
"loss": 0.036,
|
| 308 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 309 |
+
"rewards/chosen": -0.10827045142650604,
|
| 310 |
+
"rewards/margins": 0.2207660973072052,
|
| 311 |
+
"rewards/rejected": -0.32903656363487244,
|
| 312 |
+
"step": 170
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.37656903765690375,
|
| 316 |
+
"grad_norm": 3.2906223684736924,
|
| 317 |
+
"learning_rate": 3.9247834624635404e-07,
|
| 318 |
+
"logits/chosen": -2.429305076599121,
|
| 319 |
+
"logits/rejected": -2.3288915157318115,
|
| 320 |
+
"logps/chosen": -327.5008239746094,
|
| 321 |
+
"logps/rejected": -317.72308349609375,
|
| 322 |
+
"loss": 0.0369,
|
| 323 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 324 |
+
"rewards/chosen": -0.23194444179534912,
|
| 325 |
+
"rewards/margins": 0.2526394724845886,
|
| 326 |
+
"rewards/rejected": -0.48458394408226013,
|
| 327 |
+
"step": 180
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.39748953974895396,
|
| 331 |
+
"grad_norm": 3.8317278665105072,
|
| 332 |
+
"learning_rate": 3.7710310482256523e-07,
|
| 333 |
+
"logits/chosen": -2.531461000442505,
|
| 334 |
+
"logits/rejected": -2.459372043609619,
|
| 335 |
+
"logps/chosen": -271.7269287109375,
|
| 336 |
+
"logps/rejected": -283.0235595703125,
|
| 337 |
+
"loss": 0.0349,
|
| 338 |
+
"rewards/accuracies": 0.706250011920929,
|
| 339 |
+
"rewards/chosen": -0.1794736236333847,
|
| 340 |
+
"rewards/margins": 0.21512684226036072,
|
| 341 |
+
"rewards/rejected": -0.39460045099258423,
|
| 342 |
+
"step": 190
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 0.41841004184100417,
|
| 346 |
+
"grad_norm": 3.3372690013710624,
|
| 347 |
+
"learning_rate": 3.610497133404795e-07,
|
| 348 |
+
"logits/chosen": -2.586259365081787,
|
| 349 |
+
"logits/rejected": -2.531468391418457,
|
| 350 |
+
"logps/chosen": -277.77606201171875,
|
| 351 |
+
"logps/rejected": -270.3326721191406,
|
| 352 |
+
"loss": 0.0354,
|
| 353 |
+
"rewards/accuracies": 0.706250011920929,
|
| 354 |
+
"rewards/chosen": -0.17447002232074738,
|
| 355 |
+
"rewards/margins": 0.22541049122810364,
|
| 356 |
+
"rewards/rejected": -0.39988046884536743,
|
| 357 |
+
"step": 200
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.41841004184100417,
|
| 361 |
+
"eval_logits/chosen": -2.6785476207733154,
|
| 362 |
+
"eval_logits/rejected": -2.6354880332946777,
|
| 363 |
+
"eval_logps/chosen": -278.7134094238281,
|
| 364 |
+
"eval_logps/rejected": -306.79486083984375,
|
| 365 |
+
"eval_loss": 0.036840494722127914,
|
| 366 |
+
"eval_rewards/accuracies": 0.78125,
|
| 367 |
+
"eval_rewards/chosen": -0.16083484888076782,
|
| 368 |
+
"eval_rewards/margins": 0.28049272298812866,
|
| 369 |
+
"eval_rewards/rejected": -0.4413275718688965,
|
| 370 |
+
"eval_runtime": 102.7138,
|
| 371 |
+
"eval_samples_per_second": 19.472,
|
| 372 |
+
"eval_steps_per_second": 0.312,
|
| 373 |
+
"step": 200
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.4393305439330544,
|
| 377 |
+
"grad_norm": 3.528281930925339,
|
| 378 |
+
"learning_rate": 3.4440382358952115e-07,
|
| 379 |
+
"logits/chosen": -2.6202356815338135,
|
| 380 |
+
"logits/rejected": -2.496006727218628,
|
| 381 |
+
"logps/chosen": -295.03680419921875,
|
| 382 |
+
"logps/rejected": -302.34271240234375,
|
| 383 |
+
"loss": 0.0372,
|
| 384 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 385 |
+
"rewards/chosen": -0.23145723342895508,
|
| 386 |
+
"rewards/margins": 0.2113940268754959,
|
| 387 |
+
"rewards/rejected": -0.44285130500793457,
|
| 388 |
+
"step": 210
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.4602510460251046,
|
| 392 |
+
"grad_norm": 3.160346667914988,
|
| 393 |
+
"learning_rate": 3.272542485937368e-07,
|
| 394 |
+
"logits/chosen": -2.634601593017578,
|
| 395 |
+
"logits/rejected": -2.576793670654297,
|
| 396 |
+
"logps/chosen": -294.7489318847656,
|
| 397 |
+
"logps/rejected": -305.5950012207031,
|
| 398 |
+
"loss": 0.0366,
|
| 399 |
+
"rewards/accuracies": 0.75,
|
| 400 |
+
"rewards/chosen": -0.10495420545339584,
|
| 401 |
+
"rewards/margins": 0.23910054564476013,
|
| 402 |
+
"rewards/rejected": -0.3440547585487366,
|
| 403 |
+
"step": 220
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.4811715481171548,
|
| 407 |
+
"grad_norm": 3.1995588594588478,
|
| 408 |
+
"learning_rate": 3.096924887558854e-07,
|
| 409 |
+
"logits/chosen": -2.62168288230896,
|
| 410 |
+
"logits/rejected": -2.5223376750946045,
|
| 411 |
+
"logps/chosen": -295.1716003417969,
|
| 412 |
+
"logps/rejected": -300.92059326171875,
|
| 413 |
+
"loss": 0.0332,
|
| 414 |
+
"rewards/accuracies": 0.762499988079071,
|
| 415 |
+
"rewards/chosen": -0.1310446560382843,
|
| 416 |
+
"rewards/margins": 0.28820374608039856,
|
| 417 |
+
"rewards/rejected": -0.41924840211868286,
|
| 418 |
+
"step": 230
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.502092050209205,
|
| 422 |
+
"grad_norm": 4.3013803617394055,
|
| 423 |
+
"learning_rate": 2.9181224366319943e-07,
|
| 424 |
+
"logits/chosen": -2.3905320167541504,
|
| 425 |
+
"logits/rejected": -2.253870725631714,
|
| 426 |
+
"logps/chosen": -276.6922912597656,
|
| 427 |
+
"logps/rejected": -267.25909423828125,
|
| 428 |
+
"loss": 0.0359,
|
| 429 |
+
"rewards/accuracies": 0.768750011920929,
|
| 430 |
+
"rewards/chosen": -0.22041603922843933,
|
| 431 |
+
"rewards/margins": 0.2721412181854248,
|
| 432 |
+
"rewards/rejected": -0.49255722761154175,
|
| 433 |
+
"step": 240
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.5230125523012552,
|
| 437 |
+
"grad_norm": 2.9945903323401217,
|
| 438 |
+
"learning_rate": 2.7370891215954565e-07,
|
| 439 |
+
"logits/chosen": -2.575193166732788,
|
| 440 |
+
"logits/rejected": -2.45214581489563,
|
| 441 |
+
"logps/chosen": -307.0121154785156,
|
| 442 |
+
"logps/rejected": -283.763916015625,
|
| 443 |
+
"loss": 0.035,
|
| 444 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 445 |
+
"rewards/chosen": -0.1271018236875534,
|
| 446 |
+
"rewards/margins": 0.24608302116394043,
|
| 447 |
+
"rewards/rejected": -0.37318480014801025,
|
| 448 |
+
"step": 250
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 0.5230125523012552,
|
| 452 |
+
"eval_logits/chosen": -2.5931034088134766,
|
| 453 |
+
"eval_logits/rejected": -2.536372184753418,
|
| 454 |
+
"eval_logps/chosen": -265.3905029296875,
|
| 455 |
+
"eval_logps/rejected": -292.68170166015625,
|
| 456 |
+
"eval_loss": 0.03588278219103813,
|
| 457 |
+
"eval_rewards/accuracies": 0.78125,
|
| 458 |
+
"eval_rewards/chosen": -0.027605721727013588,
|
| 459 |
+
"eval_rewards/margins": 0.27259036898612976,
|
| 460 |
+
"eval_rewards/rejected": -0.3001960813999176,
|
| 461 |
+
"eval_runtime": 105.5375,
|
| 462 |
+
"eval_samples_per_second": 18.951,
|
| 463 |
+
"eval_steps_per_second": 0.303,
|
| 464 |
+
"step": 250
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.5439330543933054,
|
| 468 |
+
"grad_norm": 3.8379218102661117,
|
| 469 |
+
"learning_rate": 2.55479083351317e-07,
|
| 470 |
+
"logits/chosen": -2.571600914001465,
|
| 471 |
+
"logits/rejected": -2.4651172161102295,
|
| 472 |
+
"logps/chosen": -318.83074951171875,
|
| 473 |
+
"logps/rejected": -297.157470703125,
|
| 474 |
+
"loss": 0.0347,
|
| 475 |
+
"rewards/accuracies": 0.78125,
|
| 476 |
+
"rewards/chosen": -0.043129194527864456,
|
| 477 |
+
"rewards/margins": 0.2787154018878937,
|
| 478 |
+
"rewards/rejected": -0.32184460759162903,
|
| 479 |
+
"step": 260
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.5648535564853556,
|
| 483 |
+
"grad_norm": 2.652257544306236,
|
| 484 |
+
"learning_rate": 2.3722002126275822e-07,
|
| 485 |
+
"logits/chosen": -2.4981651306152344,
|
| 486 |
+
"logits/rejected": -2.3565025329589844,
|
| 487 |
+
"logps/chosen": -312.01361083984375,
|
| 488 |
+
"logps/rejected": -299.30072021484375,
|
| 489 |
+
"loss": 0.0342,
|
| 490 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 491 |
+
"rewards/chosen": -0.18602418899536133,
|
| 492 |
+
"rewards/margins": 0.28917396068573,
|
| 493 |
+
"rewards/rejected": -0.4751981794834137,
|
| 494 |
+
"step": 270
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.5857740585774058,
|
| 498 |
+
"grad_norm": 3.8427126345253204,
|
| 499 |
+
"learning_rate": 2.19029145890313e-07,
|
| 500 |
+
"logits/chosen": -2.4173197746276855,
|
| 501 |
+
"logits/rejected": -2.354979991912842,
|
| 502 |
+
"logps/chosen": -276.7501525878906,
|
| 503 |
+
"logps/rejected": -297.5990295410156,
|
| 504 |
+
"loss": 0.035,
|
| 505 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 506 |
+
"rewards/chosen": -0.19395743310451508,
|
| 507 |
+
"rewards/margins": 0.2593483030796051,
|
| 508 |
+
"rewards/rejected": -0.45330578088760376,
|
| 509 |
+
"step": 280
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 0.606694560669456,
|
| 513 |
+
"grad_norm": 3.3214057827825507,
|
| 514 |
+
"learning_rate": 2.0100351342479216e-07,
|
| 515 |
+
"logits/chosen": -2.464021921157837,
|
| 516 |
+
"logits/rejected": -2.39694881439209,
|
| 517 |
+
"logps/chosen": -300.59075927734375,
|
| 518 |
+
"logps/rejected": -315.4048156738281,
|
| 519 |
+
"loss": 0.0356,
|
| 520 |
+
"rewards/accuracies": 0.78125,
|
| 521 |
+
"rewards/chosen": -0.09249688684940338,
|
| 522 |
+
"rewards/margins": 0.27210110425949097,
|
| 523 |
+
"rewards/rejected": -0.36459797620773315,
|
| 524 |
+
"step": 290
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.6276150627615062,
|
| 528 |
+
"grad_norm": 3.077238623321376,
|
| 529 |
+
"learning_rate": 1.8323929841460178e-07,
|
| 530 |
+
"logits/chosen": -2.3226752281188965,
|
| 531 |
+
"logits/rejected": -2.219759225845337,
|
| 532 |
+
"logps/chosen": -283.2825622558594,
|
| 533 |
+
"logps/rejected": -318.3642578125,
|
| 534 |
+
"loss": 0.0336,
|
| 535 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 536 |
+
"rewards/chosen": -0.13956645131111145,
|
| 537 |
+
"rewards/margins": 0.2606472074985504,
|
| 538 |
+
"rewards/rejected": -0.4002136290073395,
|
| 539 |
+
"step": 300
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.6276150627615062,
|
| 543 |
+
"eval_logits/chosen": -2.4060091972351074,
|
| 544 |
+
"eval_logits/rejected": -2.3179163932800293,
|
| 545 |
+
"eval_logps/chosen": -278.71954345703125,
|
| 546 |
+
"eval_logps/rejected": -307.55657958984375,
|
| 547 |
+
"eval_loss": 0.0350707545876503,
|
| 548 |
+
"eval_rewards/accuracies": 0.7734375,
|
| 549 |
+
"eval_rewards/chosen": -0.16089613735675812,
|
| 550 |
+
"eval_rewards/margins": 0.28804877400398254,
|
| 551 |
+
"eval_rewards/rejected": -0.44894489645957947,
|
| 552 |
+
"eval_runtime": 102.9876,
|
| 553 |
+
"eval_samples_per_second": 19.42,
|
| 554 |
+
"eval_steps_per_second": 0.311,
|
| 555 |
+
"step": 300
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.6485355648535565,
|
| 559 |
+
"grad_norm": 3.9837593350930987,
|
| 560 |
+
"learning_rate": 1.6583128063291573e-07,
|
| 561 |
+
"logits/chosen": -2.4029016494750977,
|
| 562 |
+
"logits/rejected": -2.2795987129211426,
|
| 563 |
+
"logps/chosen": -311.8950500488281,
|
| 564 |
+
"logps/rejected": -322.6264953613281,
|
| 565 |
+
"loss": 0.0338,
|
| 566 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 567 |
+
"rewards/chosen": -0.17933650314807892,
|
| 568 |
+
"rewards/margins": 0.2761048972606659,
|
| 569 |
+
"rewards/rejected": -0.455441415309906,
|
| 570 |
+
"step": 310
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.6694560669456067,
|
| 574 |
+
"grad_norm": 3.564824868243538,
|
| 575 |
+
"learning_rate": 1.488723393865766e-07,
|
| 576 |
+
"logits/chosen": -2.4168436527252197,
|
| 577 |
+
"logits/rejected": -2.22420072555542,
|
| 578 |
+
"logps/chosen": -310.73724365234375,
|
| 579 |
+
"logps/rejected": -295.3639831542969,
|
| 580 |
+
"loss": 0.0331,
|
| 581 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 582 |
+
"rewards/chosen": -0.18507453799247742,
|
| 583 |
+
"rewards/margins": 0.24890851974487305,
|
| 584 |
+
"rewards/rejected": -0.43398308753967285,
|
| 585 |
+
"step": 320
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.6903765690376569,
|
| 589 |
+
"grad_norm": 3.612448436540657,
|
| 590 |
+
"learning_rate": 1.3245295796480788e-07,
|
| 591 |
+
"logits/chosen": -2.4088501930236816,
|
| 592 |
+
"logits/rejected": -2.2718417644500732,
|
| 593 |
+
"logps/chosen": -300.2972717285156,
|
| 594 |
+
"logps/rejected": -283.8664245605469,
|
| 595 |
+
"loss": 0.0331,
|
| 596 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 597 |
+
"rewards/chosen": -0.1666984260082245,
|
| 598 |
+
"rewards/margins": 0.2534824311733246,
|
| 599 |
+
"rewards/rejected": -0.42018088698387146,
|
| 600 |
+
"step": 330
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 0.7112970711297071,
|
| 604 |
+
"grad_norm": 3.369937066189534,
|
| 605 |
+
"learning_rate": 1.1666074087171627e-07,
|
| 606 |
+
"logits/chosen": -2.3149194717407227,
|
| 607 |
+
"logits/rejected": -2.21246337890625,
|
| 608 |
+
"logps/chosen": -259.3164978027344,
|
| 609 |
+
"logps/rejected": -299.25360107421875,
|
| 610 |
+
"loss": 0.0344,
|
| 611 |
+
"rewards/accuracies": 0.768750011920929,
|
| 612 |
+
"rewards/chosen": -0.1287737488746643,
|
| 613 |
+
"rewards/margins": 0.26817721128463745,
|
| 614 |
+
"rewards/rejected": -0.39695096015930176,
|
| 615 |
+
"step": 340
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.7322175732217573,
|
| 619 |
+
"grad_norm": 3.0588084178439083,
|
| 620 |
+
"learning_rate": 1.0157994641835734e-07,
|
| 621 |
+
"logits/chosen": -2.3560991287231445,
|
| 622 |
+
"logits/rejected": -2.2018752098083496,
|
| 623 |
+
"logps/chosen": -296.5240478515625,
|
| 624 |
+
"logps/rejected": -308.5161437988281,
|
| 625 |
+
"loss": 0.0338,
|
| 626 |
+
"rewards/accuracies": 0.75,
|
| 627 |
+
"rewards/chosen": -0.14297232031822205,
|
| 628 |
+
"rewards/margins": 0.2560271918773651,
|
| 629 |
+
"rewards/rejected": -0.39899951219558716,
|
| 630 |
+
"step": 350
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 0.7322175732217573,
|
| 634 |
+
"eval_logits/chosen": -2.432863712310791,
|
| 635 |
+
"eval_logits/rejected": -2.3603291511535645,
|
| 636 |
+
"eval_logps/chosen": -274.0787353515625,
|
| 637 |
+
"eval_logps/rejected": -302.0604248046875,
|
| 638 |
+
"eval_loss": 0.03481597825884819,
|
| 639 |
+
"eval_rewards/accuracies": 0.76953125,
|
| 640 |
+
"eval_rewards/chosen": -0.11448819935321808,
|
| 641 |
+
"eval_rewards/margins": 0.27949514985084534,
|
| 642 |
+
"eval_rewards/rejected": -0.3939833641052246,
|
| 643 |
+
"eval_runtime": 104.1739,
|
| 644 |
+
"eval_samples_per_second": 19.199,
|
| 645 |
+
"eval_steps_per_second": 0.307,
|
| 646 |
+
"step": 350
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.7531380753138075,
|
| 650 |
+
"grad_norm": 3.2773134868984153,
|
| 651 |
+
"learning_rate": 8.729103716819111e-08,
|
| 652 |
+
"logits/chosen": -2.4333927631378174,
|
| 653 |
+
"logits/rejected": -2.364302158355713,
|
| 654 |
+
"logps/chosen": -261.3931579589844,
|
| 655 |
+
"logps/rejected": -285.6098937988281,
|
| 656 |
+
"loss": 0.0353,
|
| 657 |
+
"rewards/accuracies": 0.762499988079071,
|
| 658 |
+
"rewards/chosen": -0.14534708857536316,
|
| 659 |
+
"rewards/margins": 0.2542579770088196,
|
| 660 |
+
"rewards/rejected": -0.39960503578186035,
|
| 661 |
+
"step": 360
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.7740585774058577,
|
| 665 |
+
"grad_norm": 4.682836628884491,
|
| 666 |
+
"learning_rate": 7.387025063449081e-08,
|
| 667 |
+
"logits/chosen": -2.5038342475891113,
|
| 668 |
+
"logits/rejected": -2.4199085235595703,
|
| 669 |
+
"logps/chosen": -296.49896240234375,
|
| 670 |
+
"logps/rejected": -309.2629699707031,
|
| 671 |
+
"loss": 0.0336,
|
| 672 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 673 |
+
"rewards/chosen": -0.0865456834435463,
|
| 674 |
+
"rewards/margins": 0.28183671832084656,
|
| 675 |
+
"rewards/rejected": -0.36838242411613464,
|
| 676 |
+
"step": 370
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.7949790794979079,
|
| 680 |
+
"grad_norm": 3.127130773812614,
|
| 681 |
+
"learning_rate": 6.138919252022435e-08,
|
| 682 |
+
"logits/chosen": -2.5170111656188965,
|
| 683 |
+
"logits/rejected": -2.4522595405578613,
|
| 684 |
+
"logps/chosen": -325.26849365234375,
|
| 685 |
+
"logps/rejected": -310.82513427734375,
|
| 686 |
+
"loss": 0.0341,
|
| 687 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 688 |
+
"rewards/chosen": -0.08804645389318466,
|
| 689 |
+
"rewards/margins": 0.19630609452724457,
|
| 690 |
+
"rewards/rejected": -0.28435254096984863,
|
| 691 |
+
"step": 380
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 0.8158995815899581,
|
| 695 |
+
"grad_norm": 3.5675081888384224,
|
| 696 |
+
"learning_rate": 4.991445467064689e-08,
|
| 697 |
+
"logits/chosen": -2.4555935859680176,
|
| 698 |
+
"logits/rejected": -2.34989595413208,
|
| 699 |
+
"logps/chosen": -316.98394775390625,
|
| 700 |
+
"logps/rejected": -330.12835693359375,
|
| 701 |
+
"loss": 0.0353,
|
| 702 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 703 |
+
"rewards/chosen": -0.10423789918422699,
|
| 704 |
+
"rewards/margins": 0.29529544711112976,
|
| 705 |
+
"rewards/rejected": -0.39953336119651794,
|
| 706 |
+
"step": 390
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 0.8368200836820083,
|
| 710 |
+
"grad_norm": 3.22998220338436,
|
| 711 |
+
"learning_rate": 3.9507259776993954e-08,
|
| 712 |
+
"logits/chosen": -2.4702091217041016,
|
| 713 |
+
"logits/rejected": -2.4040184020996094,
|
| 714 |
+
"logps/chosen": -308.82708740234375,
|
| 715 |
+
"logps/rejected": -316.28070068359375,
|
| 716 |
+
"loss": 0.0352,
|
| 717 |
+
"rewards/accuracies": 0.731249988079071,
|
| 718 |
+
"rewards/chosen": -0.14892233908176422,
|
| 719 |
+
"rewards/margins": 0.22808849811553955,
|
| 720 |
+
"rewards/rejected": -0.37701085209846497,
|
| 721 |
+
"step": 400
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 0.8368200836820083,
|
| 725 |
+
"eval_logits/chosen": -2.511061429977417,
|
| 726 |
+
"eval_logits/rejected": -2.437195062637329,
|
| 727 |
+
"eval_logps/chosen": -275.127685546875,
|
| 728 |
+
"eval_logps/rejected": -303.78619384765625,
|
| 729 |
+
"eval_loss": 0.034544143825769424,
|
| 730 |
+
"eval_rewards/accuracies": 0.7734375,
|
| 731 |
+
"eval_rewards/chosen": -0.1249774694442749,
|
| 732 |
+
"eval_rewards/margins": 0.28626346588134766,
|
| 733 |
+
"eval_rewards/rejected": -0.41124093532562256,
|
| 734 |
+
"eval_runtime": 104.7045,
|
| 735 |
+
"eval_samples_per_second": 19.101,
|
| 736 |
+
"eval_steps_per_second": 0.306,
|
| 737 |
+
"step": 400
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.8577405857740585,
|
| 741 |
+
"grad_norm": 3.3828994704557025,
|
| 742 |
+
"learning_rate": 3.022313472693447e-08,
|
| 743 |
+
"logits/chosen": -2.5727744102478027,
|
| 744 |
+
"logits/rejected": -2.472245931625366,
|
| 745 |
+
"logps/chosen": -329.69915771484375,
|
| 746 |
+
"logps/rejected": -336.42913818359375,
|
| 747 |
+
"loss": 0.0333,
|
| 748 |
+
"rewards/accuracies": 0.737500011920929,
|
| 749 |
+
"rewards/chosen": -0.1269669234752655,
|
| 750 |
+
"rewards/margins": 0.2546684741973877,
|
| 751 |
+
"rewards/rejected": -0.3816354274749756,
|
| 752 |
+
"step": 410
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.8786610878661087,
|
| 756 |
+
"grad_norm": 3.1926804873738943,
|
| 757 |
+
"learning_rate": 2.2111614344599684e-08,
|
| 758 |
+
"logits/chosen": -2.4479329586029053,
|
| 759 |
+
"logits/rejected": -2.3029239177703857,
|
| 760 |
+
"logps/chosen": -310.84954833984375,
|
| 761 |
+
"logps/rejected": -311.79693603515625,
|
| 762 |
+
"loss": 0.0314,
|
| 763 |
+
"rewards/accuracies": 0.737500011920929,
|
| 764 |
+
"rewards/chosen": -0.14661455154418945,
|
| 765 |
+
"rewards/margins": 0.2614172101020813,
|
| 766 |
+
"rewards/rejected": -0.40803179144859314,
|
| 767 |
+
"step": 420
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 0.899581589958159,
|
| 771 |
+
"grad_norm": 2.8296496249212364,
|
| 772 |
+
"learning_rate": 1.521597710086439e-08,
|
| 773 |
+
"logits/chosen": -2.5343031883239746,
|
| 774 |
+
"logits/rejected": -2.38383412361145,
|
| 775 |
+
"logps/chosen": -310.29693603515625,
|
| 776 |
+
"logps/rejected": -298.6971130371094,
|
| 777 |
+
"loss": 0.0342,
|
| 778 |
+
"rewards/accuracies": 0.731249988079071,
|
| 779 |
+
"rewards/chosen": -0.14315804839134216,
|
| 780 |
+
"rewards/margins": 0.250461608171463,
|
| 781 |
+
"rewards/rejected": -0.39361968636512756,
|
| 782 |
+
"step": 430
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 0.9205020920502092,
|
| 786 |
+
"grad_norm": 2.966812625966253,
|
| 787 |
+
"learning_rate": 9.57301420397924e-09,
|
| 788 |
+
"logits/chosen": -2.479175329208374,
|
| 789 |
+
"logits/rejected": -2.299555778503418,
|
| 790 |
+
"logps/chosen": -298.52862548828125,
|
| 791 |
+
"logps/rejected": -290.23345947265625,
|
| 792 |
+
"loss": 0.0336,
|
| 793 |
+
"rewards/accuracies": 0.71875,
|
| 794 |
+
"rewards/chosen": -0.16254766285419464,
|
| 795 |
+
"rewards/margins": 0.2794143557548523,
|
| 796 |
+
"rewards/rejected": -0.4419620633125305,
|
| 797 |
+
"step": 440
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 0.9414225941422594,
|
| 801 |
+
"grad_norm": 3.0686097215961046,
|
| 802 |
+
"learning_rate": 5.212833302556258e-09,
|
| 803 |
+
"logits/chosen": -2.508073091506958,
|
| 804 |
+
"logits/rejected": -2.4404401779174805,
|
| 805 |
+
"logps/chosen": -299.69134521484375,
|
| 806 |
+
"logps/rejected": -320.4561767578125,
|
| 807 |
+
"loss": 0.0342,
|
| 808 |
+
"rewards/accuracies": 0.793749988079071,
|
| 809 |
+
"rewards/chosen": -0.14945265650749207,
|
| 810 |
+
"rewards/margins": 0.2798132002353668,
|
| 811 |
+
"rewards/rejected": -0.4292658269405365,
|
| 812 |
+
"step": 450
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.9414225941422594,
|
| 816 |
+
"eval_logits/chosen": -2.500336170196533,
|
| 817 |
+
"eval_logits/rejected": -2.419475793838501,
|
| 818 |
+
"eval_logps/chosen": -275.03082275390625,
|
| 819 |
+
"eval_logps/rejected": -304.2409362792969,
|
| 820 |
+
"eval_loss": 0.03452227637171745,
|
| 821 |
+
"eval_rewards/accuracies": 0.7734375,
|
| 822 |
+
"eval_rewards/chosen": -0.12400892376899719,
|
| 823 |
+
"eval_rewards/margins": 0.29177966713905334,
|
| 824 |
+
"eval_rewards/rejected": -0.41578859090805054,
|
| 825 |
+
"eval_runtime": 102.9541,
|
| 826 |
+
"eval_samples_per_second": 19.426,
|
| 827 |
+
"eval_steps_per_second": 0.311,
|
| 828 |
+
"step": 450
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.9623430962343096,
|
| 832 |
+
"grad_norm": 3.2166545676628093,
|
| 833 |
+
"learning_rate": 2.158697848236607e-09,
|
| 834 |
+
"logits/chosen": -2.4364752769470215,
|
| 835 |
+
"logits/rejected": -2.3313584327697754,
|
| 836 |
+
"logps/chosen": -281.92254638671875,
|
| 837 |
+
"logps/rejected": -301.19110107421875,
|
| 838 |
+
"loss": 0.0341,
|
| 839 |
+
"rewards/accuracies": 0.731249988079071,
|
| 840 |
+
"rewards/chosen": -0.18644972145557404,
|
| 841 |
+
"rewards/margins": 0.2685782313346863,
|
| 842 |
+
"rewards/rejected": -0.4550279676914215,
|
| 843 |
+
"step": 460
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.9832635983263598,
|
| 847 |
+
"grad_norm": 2.5305189796072156,
|
| 848 |
+
"learning_rate": 4.269029751107489e-10,
|
| 849 |
+
"logits/chosen": -2.399737596511841,
|
| 850 |
+
"logits/rejected": -2.354271650314331,
|
| 851 |
+
"logps/chosen": -275.69482421875,
|
| 852 |
+
"logps/rejected": -305.2413635253906,
|
| 853 |
+
"loss": 0.0347,
|
| 854 |
+
"rewards/accuracies": 0.737500011920929,
|
| 855 |
+
"rewards/chosen": -0.12569288909435272,
|
| 856 |
+
"rewards/margins": 0.26325756311416626,
|
| 857 |
+
"rewards/rejected": -0.38895049691200256,
|
| 858 |
+
"step": 470
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 1.0,
|
| 862 |
+
"step": 478,
|
| 863 |
+
"total_flos": 0.0,
|
| 864 |
+
"train_loss": 0.037969033089641745,
|
| 865 |
+
"train_runtime": 12745.9722,
|
| 866 |
+
"train_samples_per_second": 4.796,
|
| 867 |
+
"train_steps_per_second": 0.038
|
| 868 |
+
}
|
| 869 |
+
],
|
| 870 |
+
"logging_steps": 10,
|
| 871 |
+
"max_steps": 478,
|
| 872 |
+
"num_input_tokens_seen": 0,
|
| 873 |
+
"num_train_epochs": 1,
|
| 874 |
+
"save_steps": 100,
|
| 875 |
+
"stateful_callbacks": {
|
| 876 |
+
"TrainerControl": {
|
| 877 |
+
"args": {
|
| 878 |
+
"should_epoch_stop": false,
|
| 879 |
+
"should_evaluate": false,
|
| 880 |
+
"should_log": false,
|
| 881 |
+
"should_save": true,
|
| 882 |
+
"should_training_stop": true
|
| 883 |
+
},
|
| 884 |
+
"attributes": {}
|
| 885 |
+
}
|
| 886 |
+
},
|
| 887 |
+
"total_flos": 0.0,
|
| 888 |
+
"train_batch_size": 8,
|
| 889 |
+
"trial_name": null,
|
| 890 |
+
"trial_params": null
|
| 891 |
+
}
|