Model save
Browse files- README.md +75 -0
- all_results.json +21 -0
- eval_results.json +16 -0
- generation_config.json +6 -0
- model.safetensors +1 -1
- runs/Mar05_13-08-21_u124757/events.out.tfevents.1709673156.u124757.331914.0 +2 -2
- runs/Mar05_13-08-21_u124757/events.out.tfevents.1709676893.u124757.331914.1 +3 -0
- train_results.json +8 -0
- trainer_state.json +766 -0
README.md
ADDED
|
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: btherien/410M_it-132366_Tr-slim-pajama-300B-replay5_finetune_sft-full
|
| 3 |
+
tags:
|
| 4 |
+
- trl
|
| 5 |
+
- dpo
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
model-index:
|
| 8 |
+
- name: 410M_it-132366_Tr-slim-pajama-300B-replay5_finetune_dpo-full
|
| 9 |
+
results: []
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 13 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 14 |
+
|
| 15 |
+
# 410M_it-132366_Tr-slim-pajama-300B-replay5_finetune_dpo-full
|
| 16 |
+
|
| 17 |
+
This model is a fine-tuned version of [btherien/410M_it-132366_Tr-slim-pajama-300B-replay5_finetune_sft-full](https://huggingface.co/btherien/410M_it-132366_Tr-slim-pajama-300B-replay5_finetune_sft-full) on the None dataset.
|
| 18 |
+
It achieves the following results on the evaluation set:
|
| 19 |
+
- Loss: 0.6500
|
| 20 |
+
- Rewards/chosen: -0.4023
|
| 21 |
+
- Rewards/rejected: -0.5547
|
| 22 |
+
- Rewards/accuracies: 0.6171
|
| 23 |
+
- Rewards/margins: 0.1514
|
| 24 |
+
- Logps/rejected: -420.0
|
| 25 |
+
- Logps/chosen: -464.0
|
| 26 |
+
- Logits/rejected: -1.3828
|
| 27 |
+
- Logits/chosen: -1.4297
|
| 28 |
+
|
| 29 |
+
## Model description
|
| 30 |
+
|
| 31 |
+
More information needed
|
| 32 |
+
|
| 33 |
+
## Intended uses & limitations
|
| 34 |
+
|
| 35 |
+
More information needed
|
| 36 |
+
|
| 37 |
+
## Training and evaluation data
|
| 38 |
+
|
| 39 |
+
More information needed
|
| 40 |
+
|
| 41 |
+
## Training procedure
|
| 42 |
+
|
| 43 |
+
### Training hyperparameters
|
| 44 |
+
|
| 45 |
+
The following hyperparameters were used during training:
|
| 46 |
+
- learning_rate: 5e-07
|
| 47 |
+
- train_batch_size: 8
|
| 48 |
+
- eval_batch_size: 8
|
| 49 |
+
- seed: 42
|
| 50 |
+
- distributed_type: multi-GPU
|
| 51 |
+
- num_devices: 4
|
| 52 |
+
- gradient_accumulation_steps: 4
|
| 53 |
+
- total_train_batch_size: 128
|
| 54 |
+
- total_eval_batch_size: 32
|
| 55 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 56 |
+
- lr_scheduler_type: cosine
|
| 57 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 58 |
+
- num_epochs: 1
|
| 59 |
+
|
| 60 |
+
### Training results
|
| 61 |
+
|
| 62 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 63 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 64 |
+
| 0.6867 | 0.21 | 100 | 0.6853 | -0.0056 | -0.0271 | 0.5556 | 0.0216 | -368.0 | -424.0 | -1.1953 | -1.2422 |
|
| 65 |
+
| 0.6632 | 0.42 | 200 | 0.6625 | -0.2285 | -0.3203 | 0.625 | 0.0913 | -396.0 | -446.0 | -1.3203 | -1.3672 |
|
| 66 |
+
| 0.6493 | 0.63 | 300 | 0.6532 | -0.3555 | -0.4902 | 0.625 | 0.1348 | -412.0 | -460.0 | -1.3516 | -1.3984 |
|
| 67 |
+
| 0.6462 | 0.84 | 400 | 0.6495 | -0.4004 | -0.5547 | 0.6389 | 0.1553 | -420.0 | -464.0 | -1.375 | -1.4219 |
|
| 68 |
+
|
| 69 |
+
|
| 70 |
+
### Framework versions
|
| 71 |
+
|
| 72 |
+
- Transformers 4.36.2
|
| 73 |
+
- Pytorch 2.0.1+cu117
|
| 74 |
+
- Datasets 2.14.6
|
| 75 |
+
- Tokenizers 0.15.2
|
all_results.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_logits/chosen": -1.4296875,
|
| 4 |
+
"eval_logits/rejected": -1.3828125,
|
| 5 |
+
"eval_logps/chosen": -464.0,
|
| 6 |
+
"eval_logps/rejected": -420.0,
|
| 7 |
+
"eval_loss": 0.6500468850135803,
|
| 8 |
+
"eval_rewards/accuracies": 0.6170634627342224,
|
| 9 |
+
"eval_rewards/chosen": -0.40234375,
|
| 10 |
+
"eval_rewards/margins": 0.1513671875,
|
| 11 |
+
"eval_rewards/rejected": -0.5546875,
|
| 12 |
+
"eval_runtime": 52.2116,
|
| 13 |
+
"eval_samples": 2000,
|
| 14 |
+
"eval_samples_per_second": 38.306,
|
| 15 |
+
"eval_steps_per_second": 1.207,
|
| 16 |
+
"train_loss": 0.6625334221861897,
|
| 17 |
+
"train_runtime": 3684.7873,
|
| 18 |
+
"train_samples": 61135,
|
| 19 |
+
"train_samples_per_second": 16.591,
|
| 20 |
+
"train_steps_per_second": 0.129
|
| 21 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_logits/chosen": -1.4296875,
|
| 4 |
+
"eval_logits/rejected": -1.3828125,
|
| 5 |
+
"eval_logps/chosen": -464.0,
|
| 6 |
+
"eval_logps/rejected": -420.0,
|
| 7 |
+
"eval_loss": 0.6500468850135803,
|
| 8 |
+
"eval_rewards/accuracies": 0.6170634627342224,
|
| 9 |
+
"eval_rewards/chosen": -0.40234375,
|
| 10 |
+
"eval_rewards/margins": 0.1513671875,
|
| 11 |
+
"eval_rewards/rejected": -0.5546875,
|
| 12 |
+
"eval_runtime": 52.2116,
|
| 13 |
+
"eval_samples": 2000,
|
| 14 |
+
"eval_samples_per_second": 38.306,
|
| 15 |
+
"eval_steps_per_second": 1.207
|
| 16 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 0,
|
| 4 |
+
"eos_token_id": 0,
|
| 5 |
+
"transformers_version": "4.36.2"
|
| 6 |
+
}
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 810702192
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c950b37cfaf9ca6200ddf854aaadf4209162d3815229ce993a50bfe535102b63
|
| 3 |
size 810702192
|
runs/Mar05_13-08-21_u124757/events.out.tfevents.1709673156.u124757.331914.0
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 sha256:9036f83ba7cb8b28ec6e7a2139264780af9b8f6fd70d8e3055b4e83ff5458862
|
| 3 |
+
size 38378
|
runs/Mar05_13-08-21_u124757/events.out.tfevents.1709676893.u124757.331914.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b31053bf0be366a280df5e4b42edf2bf15645a51a1043978676362698508e821
|
| 3 |
+
size 828
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.6625334221861897,
|
| 4 |
+
"train_runtime": 3684.7873,
|
| 5 |
+
"train_samples": 61135,
|
| 6 |
+
"train_samples_per_second": 16.591,
|
| 7 |
+
"train_steps_per_second": 0.129
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,766 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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 |
+
"learning_rate": 1.0416666666666666e-08,
|
| 14 |
+
"logits/chosen": -1.375,
|
| 15 |
+
"logits/rejected": -1.21875,
|
| 16 |
+
"logps/chosen": -404.0,
|
| 17 |
+
"logps/rejected": -374.0,
|
| 18 |
+
"loss": 0.6914,
|
| 19 |
+
"rewards/accuracies": 0.0,
|
| 20 |
+
"rewards/chosen": 0.0,
|
| 21 |
+
"rewards/margins": 0.0,
|
| 22 |
+
"rewards/rejected": 0.0,
|
| 23 |
+
"step": 1
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.02,
|
| 27 |
+
"learning_rate": 1.0416666666666667e-07,
|
| 28 |
+
"logits/chosen": -1.1875,
|
| 29 |
+
"logits/rejected": -1.15625,
|
| 30 |
+
"logps/chosen": -394.0,
|
| 31 |
+
"logps/rejected": -334.0,
|
| 32 |
+
"loss": 0.6944,
|
| 33 |
+
"rewards/accuracies": 0.34375,
|
| 34 |
+
"rewards/chosen": -0.0024261474609375,
|
| 35 |
+
"rewards/margins": -0.004364013671875,
|
| 36 |
+
"rewards/rejected": 0.001922607421875,
|
| 37 |
+
"step": 10
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.04,
|
| 41 |
+
"learning_rate": 2.0833333333333333e-07,
|
| 42 |
+
"logits/chosen": -1.2109375,
|
| 43 |
+
"logits/rejected": -1.171875,
|
| 44 |
+
"logps/chosen": -412.0,
|
| 45 |
+
"logps/rejected": -342.0,
|
| 46 |
+
"loss": 0.6931,
|
| 47 |
+
"rewards/accuracies": 0.47187501192092896,
|
| 48 |
+
"rewards/chosen": 0.000743865966796875,
|
| 49 |
+
"rewards/margins": 0.003692626953125,
|
| 50 |
+
"rewards/rejected": -0.0029449462890625,
|
| 51 |
+
"step": 20
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.06,
|
| 55 |
+
"learning_rate": 3.1249999999999997e-07,
|
| 56 |
+
"logits/chosen": -1.2109375,
|
| 57 |
+
"logits/rejected": -1.2265625,
|
| 58 |
+
"logps/chosen": -432.0,
|
| 59 |
+
"logps/rejected": -372.0,
|
| 60 |
+
"loss": 0.6923,
|
| 61 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 62 |
+
"rewards/chosen": 0.002197265625,
|
| 63 |
+
"rewards/margins": 0.003204345703125,
|
| 64 |
+
"rewards/rejected": -0.0009918212890625,
|
| 65 |
+
"step": 30
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.08,
|
| 69 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 70 |
+
"logits/chosen": -1.2421875,
|
| 71 |
+
"logits/rejected": -1.21875,
|
| 72 |
+
"logps/chosen": -444.0,
|
| 73 |
+
"logps/rejected": -368.0,
|
| 74 |
+
"loss": 0.6933,
|
| 75 |
+
"rewards/accuracies": 0.46875,
|
| 76 |
+
"rewards/chosen": -0.0019073486328125,
|
| 77 |
+
"rewards/margins": 9.357929229736328e-06,
|
| 78 |
+
"rewards/rejected": -0.00191497802734375,
|
| 79 |
+
"step": 40
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.1,
|
| 83 |
+
"learning_rate": 4.999731868769026e-07,
|
| 84 |
+
"logits/chosen": -1.1953125,
|
| 85 |
+
"logits/rejected": -1.15625,
|
| 86 |
+
"logps/chosen": -432.0,
|
| 87 |
+
"logps/rejected": -384.0,
|
| 88 |
+
"loss": 0.6918,
|
| 89 |
+
"rewards/accuracies": 0.4906249940395355,
|
| 90 |
+
"rewards/chosen": 0.000812530517578125,
|
| 91 |
+
"rewards/margins": 0.00439453125,
|
| 92 |
+
"rewards/rejected": -0.0035858154296875,
|
| 93 |
+
"step": 50
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.13,
|
| 97 |
+
"learning_rate": 4.990353313429303e-07,
|
| 98 |
+
"logits/chosen": -1.1484375,
|
| 99 |
+
"logits/rejected": -1.078125,
|
| 100 |
+
"logps/chosen": -400.0,
|
| 101 |
+
"logps/rejected": -308.0,
|
| 102 |
+
"loss": 0.6915,
|
| 103 |
+
"rewards/accuracies": 0.484375,
|
| 104 |
+
"rewards/chosen": 0.0038299560546875,
|
| 105 |
+
"rewards/margins": 0.00457763671875,
|
| 106 |
+
"rewards/rejected": -0.000759124755859375,
|
| 107 |
+
"step": 60
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.15,
|
| 111 |
+
"learning_rate": 4.967625656594781e-07,
|
| 112 |
+
"logits/chosen": -1.203125,
|
| 113 |
+
"logits/rejected": -1.203125,
|
| 114 |
+
"logps/chosen": -406.0,
|
| 115 |
+
"logps/rejected": -336.0,
|
| 116 |
+
"loss": 0.6907,
|
| 117 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 118 |
+
"rewards/chosen": -0.00010776519775390625,
|
| 119 |
+
"rewards/margins": 0.002593994140625,
|
| 120 |
+
"rewards/rejected": -0.002685546875,
|
| 121 |
+
"step": 70
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.17,
|
| 125 |
+
"learning_rate": 4.93167072587771e-07,
|
| 126 |
+
"logits/chosen": -1.2265625,
|
| 127 |
+
"logits/rejected": -1.2265625,
|
| 128 |
+
"logps/chosen": -416.0,
|
| 129 |
+
"logps/rejected": -394.0,
|
| 130 |
+
"loss": 0.6907,
|
| 131 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 132 |
+
"rewards/chosen": -0.000720977783203125,
|
| 133 |
+
"rewards/margins": 0.00144195556640625,
|
| 134 |
+
"rewards/rejected": -0.002166748046875,
|
| 135 |
+
"step": 80
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.19,
|
| 139 |
+
"learning_rate": 4.882681251368548e-07,
|
| 140 |
+
"logits/chosen": -1.171875,
|
| 141 |
+
"logits/rejected": -1.1796875,
|
| 142 |
+
"logps/chosen": -382.0,
|
| 143 |
+
"logps/rejected": -324.0,
|
| 144 |
+
"loss": 0.6871,
|
| 145 |
+
"rewards/accuracies": 0.546875,
|
| 146 |
+
"rewards/chosen": 0.0035858154296875,
|
| 147 |
+
"rewards/margins": 0.01318359375,
|
| 148 |
+
"rewards/rejected": -0.0096435546875,
|
| 149 |
+
"step": 90
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.21,
|
| 153 |
+
"learning_rate": 4.820919832540181e-07,
|
| 154 |
+
"logits/chosen": -1.28125,
|
| 155 |
+
"logits/rejected": -1.171875,
|
| 156 |
+
"logps/chosen": -420.0,
|
| 157 |
+
"logps/rejected": -372.0,
|
| 158 |
+
"loss": 0.6867,
|
| 159 |
+
"rewards/accuracies": 0.5625,
|
| 160 |
+
"rewards/chosen": -0.00103759765625,
|
| 161 |
+
"rewards/margins": 0.015625,
|
| 162 |
+
"rewards/rejected": -0.0167236328125,
|
| 163 |
+
"step": 100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.21,
|
| 167 |
+
"eval_logits/chosen": -1.2421875,
|
| 168 |
+
"eval_logits/rejected": -1.1953125,
|
| 169 |
+
"eval_logps/chosen": -424.0,
|
| 170 |
+
"eval_logps/rejected": -368.0,
|
| 171 |
+
"eval_loss": 0.6852812767028809,
|
| 172 |
+
"eval_rewards/accuracies": 0.5555555820465088,
|
| 173 |
+
"eval_rewards/chosen": -0.005645751953125,
|
| 174 |
+
"eval_rewards/margins": 0.0216064453125,
|
| 175 |
+
"eval_rewards/rejected": -0.027099609375,
|
| 176 |
+
"eval_runtime": 51.3914,
|
| 177 |
+
"eval_samples_per_second": 38.917,
|
| 178 |
+
"eval_steps_per_second": 1.226,
|
| 179 |
+
"step": 100
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.23,
|
| 183 |
+
"learning_rate": 4.7467175306295647e-07,
|
| 184 |
+
"logits/chosen": -1.234375,
|
| 185 |
+
"logits/rejected": -1.21875,
|
| 186 |
+
"logps/chosen": -448.0,
|
| 187 |
+
"logps/rejected": -366.0,
|
| 188 |
+
"loss": 0.6836,
|
| 189 |
+
"rewards/accuracies": 0.5625,
|
| 190 |
+
"rewards/chosen": -0.0045166015625,
|
| 191 |
+
"rewards/margins": 0.0247802734375,
|
| 192 |
+
"rewards/rejected": -0.029296875,
|
| 193 |
+
"step": 110
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.25,
|
| 197 |
+
"learning_rate": 4.6604720940421207e-07,
|
| 198 |
+
"logits/chosen": -1.203125,
|
| 199 |
+
"logits/rejected": -1.140625,
|
| 200 |
+
"logps/chosen": -410.0,
|
| 201 |
+
"logps/rejected": -332.0,
|
| 202 |
+
"loss": 0.6802,
|
| 203 |
+
"rewards/accuracies": 0.578125,
|
| 204 |
+
"rewards/chosen": -0.0155029296875,
|
| 205 |
+
"rewards/margins": 0.025390625,
|
| 206 |
+
"rewards/rejected": -0.040771484375,
|
| 207 |
+
"step": 120
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.27,
|
| 211 |
+
"learning_rate": 4.5626458262912735e-07,
|
| 212 |
+
"logits/chosen": -1.265625,
|
| 213 |
+
"logits/rejected": -1.2265625,
|
| 214 |
+
"logps/chosen": -404.0,
|
| 215 |
+
"logps/rejected": -380.0,
|
| 216 |
+
"loss": 0.6806,
|
| 217 |
+
"rewards/accuracies": 0.5718749761581421,
|
| 218 |
+
"rewards/chosen": -0.04638671875,
|
| 219 |
+
"rewards/margins": 0.0257568359375,
|
| 220 |
+
"rewards/rejected": -0.072265625,
|
| 221 |
+
"step": 130
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.29,
|
| 225 |
+
"learning_rate": 4.453763107901675e-07,
|
| 226 |
+
"logits/chosen": -1.265625,
|
| 227 |
+
"logits/rejected": -1.2578125,
|
| 228 |
+
"logps/chosen": -454.0,
|
| 229 |
+
"logps/rejected": -388.0,
|
| 230 |
+
"loss": 0.6776,
|
| 231 |
+
"rewards/accuracies": 0.574999988079071,
|
| 232 |
+
"rewards/chosen": -0.0654296875,
|
| 233 |
+
"rewards/margins": 0.037841796875,
|
| 234 |
+
"rewards/rejected": -0.10302734375,
|
| 235 |
+
"step": 140
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.31,
|
| 239 |
+
"learning_rate": 4.3344075855595097e-07,
|
| 240 |
+
"logits/chosen": -1.2421875,
|
| 241 |
+
"logits/rejected": -1.234375,
|
| 242 |
+
"logps/chosen": -420.0,
|
| 243 |
+
"logps/rejected": -356.0,
|
| 244 |
+
"loss": 0.6716,
|
| 245 |
+
"rewards/accuracies": 0.609375,
|
| 246 |
+
"rewards/chosen": -0.09423828125,
|
| 247 |
+
"rewards/margins": 0.049560546875,
|
| 248 |
+
"rewards/rejected": -0.1435546875,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.33,
|
| 253 |
+
"learning_rate": 4.2052190435769554e-07,
|
| 254 |
+
"logits/chosen": -1.28125,
|
| 255 |
+
"logits/rejected": -1.234375,
|
| 256 |
+
"logps/chosen": -422.0,
|
| 257 |
+
"logps/rejected": -388.0,
|
| 258 |
+
"loss": 0.6672,
|
| 259 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 260 |
+
"rewards/chosen": -0.12353515625,
|
| 261 |
+
"rewards/margins": 0.061767578125,
|
| 262 |
+
"rewards/rejected": -0.185546875,
|
| 263 |
+
"step": 160
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.36,
|
| 267 |
+
"learning_rate": 4.0668899744407567e-07,
|
| 268 |
+
"logits/chosen": -1.28125,
|
| 269 |
+
"logits/rejected": -1.2265625,
|
| 270 |
+
"logps/chosen": -390.0,
|
| 271 |
+
"logps/rejected": -336.0,
|
| 272 |
+
"loss": 0.674,
|
| 273 |
+
"rewards/accuracies": 0.59375,
|
| 274 |
+
"rewards/chosen": -0.1630859375,
|
| 275 |
+
"rewards/margins": 0.037353515625,
|
| 276 |
+
"rewards/rejected": -0.2001953125,
|
| 277 |
+
"step": 170
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.38,
|
| 281 |
+
"learning_rate": 3.920161866827889e-07,
|
| 282 |
+
"logits/chosen": -1.3125,
|
| 283 |
+
"logits/rejected": -1.296875,
|
| 284 |
+
"logps/chosen": -426.0,
|
| 285 |
+
"logps/rejected": -382.0,
|
| 286 |
+
"loss": 0.6659,
|
| 287 |
+
"rewards/accuracies": 0.574999988079071,
|
| 288 |
+
"rewards/chosen": -0.181640625,
|
| 289 |
+
"rewards/margins": 0.06298828125,
|
| 290 |
+
"rewards/rejected": -0.244140625,
|
| 291 |
+
"step": 180
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.4,
|
| 295 |
+
"learning_rate": 3.765821230985757e-07,
|
| 296 |
+
"logits/chosen": -1.3203125,
|
| 297 |
+
"logits/rejected": -1.265625,
|
| 298 |
+
"logps/chosen": -410.0,
|
| 299 |
+
"logps/rejected": -380.0,
|
| 300 |
+
"loss": 0.6594,
|
| 301 |
+
"rewards/accuracies": 0.659375011920929,
|
| 302 |
+
"rewards/chosen": -0.19140625,
|
| 303 |
+
"rewards/margins": 0.09814453125,
|
| 304 |
+
"rewards/rejected": -0.2890625,
|
| 305 |
+
"step": 190
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.42,
|
| 309 |
+
"learning_rate": 3.604695382782159e-07,
|
| 310 |
+
"logits/chosen": -1.3671875,
|
| 311 |
+
"logits/rejected": -1.265625,
|
| 312 |
+
"logps/chosen": -438.0,
|
| 313 |
+
"logps/rejected": -374.0,
|
| 314 |
+
"loss": 0.6632,
|
| 315 |
+
"rewards/accuracies": 0.621874988079071,
|
| 316 |
+
"rewards/chosen": -0.21875,
|
| 317 |
+
"rewards/margins": 0.07421875,
|
| 318 |
+
"rewards/rejected": -0.29296875,
|
| 319 |
+
"step": 200
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.42,
|
| 323 |
+
"eval_logits/chosen": -1.3671875,
|
| 324 |
+
"eval_logits/rejected": -1.3203125,
|
| 325 |
+
"eval_logps/chosen": -446.0,
|
| 326 |
+
"eval_logps/rejected": -396.0,
|
| 327 |
+
"eval_loss": 0.6625468730926514,
|
| 328 |
+
"eval_rewards/accuracies": 0.625,
|
| 329 |
+
"eval_rewards/chosen": -0.228515625,
|
| 330 |
+
"eval_rewards/margins": 0.09130859375,
|
| 331 |
+
"eval_rewards/rejected": -0.3203125,
|
| 332 |
+
"eval_runtime": 52.3612,
|
| 333 |
+
"eval_samples_per_second": 38.196,
|
| 334 |
+
"eval_steps_per_second": 1.203,
|
| 335 |
+
"step": 200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 0.44,
|
| 339 |
+
"learning_rate": 3.4376480090239047e-07,
|
| 340 |
+
"logits/chosen": -1.4296875,
|
| 341 |
+
"logits/rejected": -1.359375,
|
| 342 |
+
"logps/chosen": -442.0,
|
| 343 |
+
"logps/rejected": -388.0,
|
| 344 |
+
"loss": 0.6604,
|
| 345 |
+
"rewards/accuracies": 0.609375,
|
| 346 |
+
"rewards/chosen": -0.2421875,
|
| 347 |
+
"rewards/margins": 0.0888671875,
|
| 348 |
+
"rewards/rejected": -0.33203125,
|
| 349 |
+
"step": 210
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 0.46,
|
| 353 |
+
"learning_rate": 3.265574537815398e-07,
|
| 354 |
+
"logits/chosen": -1.296875,
|
| 355 |
+
"logits/rejected": -1.28125,
|
| 356 |
+
"logps/chosen": -406.0,
|
| 357 |
+
"logps/rejected": -374.0,
|
| 358 |
+
"loss": 0.6588,
|
| 359 |
+
"rewards/accuracies": 0.590624988079071,
|
| 360 |
+
"rewards/chosen": -0.271484375,
|
| 361 |
+
"rewards/margins": 0.0673828125,
|
| 362 |
+
"rewards/rejected": -0.33984375,
|
| 363 |
+
"step": 220
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.48,
|
| 367 |
+
"learning_rate": 3.0893973387735683e-07,
|
| 368 |
+
"logits/chosen": -1.34375,
|
| 369 |
+
"logits/rejected": -1.296875,
|
| 370 |
+
"logps/chosen": -406.0,
|
| 371 |
+
"logps/rejected": -372.0,
|
| 372 |
+
"loss": 0.6516,
|
| 373 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 374 |
+
"rewards/chosen": -0.26171875,
|
| 375 |
+
"rewards/margins": 0.1064453125,
|
| 376 |
+
"rewards/rejected": -0.369140625,
|
| 377 |
+
"step": 230
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.5,
|
| 381 |
+
"learning_rate": 2.910060778827554e-07,
|
| 382 |
+
"logits/chosen": -1.3515625,
|
| 383 |
+
"logits/rejected": -1.328125,
|
| 384 |
+
"logps/chosen": -420.0,
|
| 385 |
+
"logps/rejected": -368.0,
|
| 386 |
+
"loss": 0.6635,
|
| 387 |
+
"rewards/accuracies": 0.609375,
|
| 388 |
+
"rewards/chosen": -0.302734375,
|
| 389 |
+
"rewards/margins": 0.09619140625,
|
| 390 |
+
"rewards/rejected": -0.400390625,
|
| 391 |
+
"step": 240
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.52,
|
| 395 |
+
"learning_rate": 2.7285261601056697e-07,
|
| 396 |
+
"logits/chosen": -1.3828125,
|
| 397 |
+
"logits/rejected": -1.328125,
|
| 398 |
+
"logps/chosen": -426.0,
|
| 399 |
+
"logps/rejected": -384.0,
|
| 400 |
+
"loss": 0.6546,
|
| 401 |
+
"rewards/accuracies": 0.550000011920929,
|
| 402 |
+
"rewards/chosen": -0.30859375,
|
| 403 |
+
"rewards/margins": 0.10693359375,
|
| 404 |
+
"rewards/rejected": -0.416015625,
|
| 405 |
+
"step": 250
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.54,
|
| 409 |
+
"learning_rate": 2.5457665670441937e-07,
|
| 410 |
+
"logits/chosen": -1.3515625,
|
| 411 |
+
"logits/rejected": -1.2890625,
|
| 412 |
+
"logps/chosen": -432.0,
|
| 413 |
+
"logps/rejected": -384.0,
|
| 414 |
+
"loss": 0.6521,
|
| 415 |
+
"rewards/accuracies": 0.637499988079071,
|
| 416 |
+
"rewards/chosen": -0.337890625,
|
| 417 |
+
"rewards/margins": 0.142578125,
|
| 418 |
+
"rewards/rejected": -0.48046875,
|
| 419 |
+
"step": 260
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 0.57,
|
| 423 |
+
"learning_rate": 2.3627616503391812e-07,
|
| 424 |
+
"logits/chosen": -1.3828125,
|
| 425 |
+
"logits/rejected": -1.390625,
|
| 426 |
+
"logps/chosen": -454.0,
|
| 427 |
+
"logps/rejected": -426.0,
|
| 428 |
+
"loss": 0.6522,
|
| 429 |
+
"rewards/accuracies": 0.609375,
|
| 430 |
+
"rewards/chosen": -0.34375,
|
| 431 |
+
"rewards/margins": 0.11181640625,
|
| 432 |
+
"rewards/rejected": -0.45703125,
|
| 433 |
+
"step": 270
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.59,
|
| 437 |
+
"learning_rate": 2.1804923757009882e-07,
|
| 438 |
+
"logits/chosen": -1.296875,
|
| 439 |
+
"logits/rejected": -1.3046875,
|
| 440 |
+
"logps/chosen": -426.0,
|
| 441 |
+
"logps/rejected": -372.0,
|
| 442 |
+
"loss": 0.6545,
|
| 443 |
+
"rewards/accuracies": 0.606249988079071,
|
| 444 |
+
"rewards/chosen": -0.35546875,
|
| 445 |
+
"rewards/margins": 0.134765625,
|
| 446 |
+
"rewards/rejected": -0.48828125,
|
| 447 |
+
"step": 280
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.61,
|
| 451 |
+
"learning_rate": 1.9999357655598891e-07,
|
| 452 |
+
"logits/chosen": -1.3515625,
|
| 453 |
+
"logits/rejected": -1.3671875,
|
| 454 |
+
"logps/chosen": -436.0,
|
| 455 |
+
"logps/rejected": -388.0,
|
| 456 |
+
"loss": 0.6546,
|
| 457 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 458 |
+
"rewards/chosen": -0.37890625,
|
| 459 |
+
"rewards/margins": 0.109375,
|
| 460 |
+
"rewards/rejected": -0.48828125,
|
| 461 |
+
"step": 290
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.63,
|
| 465 |
+
"learning_rate": 1.8220596619089573e-07,
|
| 466 |
+
"logits/chosen": -1.375,
|
| 467 |
+
"logits/rejected": -1.3046875,
|
| 468 |
+
"logps/chosen": -474.0,
|
| 469 |
+
"logps/rejected": -442.0,
|
| 470 |
+
"loss": 0.6493,
|
| 471 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 472 |
+
"rewards/chosen": -0.33984375,
|
| 473 |
+
"rewards/margins": 0.12109375,
|
| 474 |
+
"rewards/rejected": -0.458984375,
|
| 475 |
+
"step": 300
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.63,
|
| 479 |
+
"eval_logits/chosen": -1.3984375,
|
| 480 |
+
"eval_logits/rejected": -1.3515625,
|
| 481 |
+
"eval_logps/chosen": -460.0,
|
| 482 |
+
"eval_logps/rejected": -412.0,
|
| 483 |
+
"eval_loss": 0.653249979019165,
|
| 484 |
+
"eval_rewards/accuracies": 0.625,
|
| 485 |
+
"eval_rewards/chosen": -0.35546875,
|
| 486 |
+
"eval_rewards/margins": 0.134765625,
|
| 487 |
+
"eval_rewards/rejected": -0.490234375,
|
| 488 |
+
"eval_runtime": 52.3802,
|
| 489 |
+
"eval_samples_per_second": 38.182,
|
| 490 |
+
"eval_steps_per_second": 1.203,
|
| 491 |
+
"step": 300
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.65,
|
| 495 |
+
"learning_rate": 1.647817538357072e-07,
|
| 496 |
+
"logits/chosen": -1.390625,
|
| 497 |
+
"logits/rejected": -1.3203125,
|
| 498 |
+
"logps/chosen": -432.0,
|
| 499 |
+
"logps/rejected": -366.0,
|
| 500 |
+
"loss": 0.6467,
|
| 501 |
+
"rewards/accuracies": 0.637499988079071,
|
| 502 |
+
"rewards/chosen": -0.353515625,
|
| 503 |
+
"rewards/margins": 0.125,
|
| 504 |
+
"rewards/rejected": -0.478515625,
|
| 505 |
+
"step": 310
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.67,
|
| 509 |
+
"learning_rate": 1.478143389201113e-07,
|
| 510 |
+
"logits/chosen": -1.296875,
|
| 511 |
+
"logits/rejected": -1.328125,
|
| 512 |
+
"logps/chosen": -416.0,
|
| 513 |
+
"logps/rejected": -386.0,
|
| 514 |
+
"loss": 0.6499,
|
| 515 |
+
"rewards/accuracies": 0.6343749761581421,
|
| 516 |
+
"rewards/chosen": -0.3515625,
|
| 517 |
+
"rewards/margins": 0.11865234375,
|
| 518 |
+
"rewards/rejected": -0.47265625,
|
| 519 |
+
"step": 320
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.69,
|
| 523 |
+
"learning_rate": 1.3139467229135998e-07,
|
| 524 |
+
"logits/chosen": -1.4296875,
|
| 525 |
+
"logits/rejected": -1.390625,
|
| 526 |
+
"logps/chosen": -476.0,
|
| 527 |
+
"logps/rejected": -422.0,
|
| 528 |
+
"loss": 0.6409,
|
| 529 |
+
"rewards/accuracies": 0.6343749761581421,
|
| 530 |
+
"rewards/chosen": -0.373046875,
|
| 531 |
+
"rewards/margins": 0.1669921875,
|
| 532 |
+
"rewards/rejected": -0.5390625,
|
| 533 |
+
"step": 330
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.71,
|
| 537 |
+
"learning_rate": 1.1561076868822755e-07,
|
| 538 |
+
"logits/chosen": -1.3671875,
|
| 539 |
+
"logits/rejected": -1.328125,
|
| 540 |
+
"logps/chosen": -428.0,
|
| 541 |
+
"logps/rejected": -404.0,
|
| 542 |
+
"loss": 0.6505,
|
| 543 |
+
"rewards/accuracies": 0.59375,
|
| 544 |
+
"rewards/chosen": -0.384765625,
|
| 545 |
+
"rewards/margins": 0.10400390625,
|
| 546 |
+
"rewards/rejected": -0.48828125,
|
| 547 |
+
"step": 340
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 0.73,
|
| 551 |
+
"learning_rate": 1.0054723495346482e-07,
|
| 552 |
+
"logits/chosen": -1.359375,
|
| 553 |
+
"logits/rejected": -1.2890625,
|
| 554 |
+
"logps/chosen": -420.0,
|
| 555 |
+
"logps/rejected": -388.0,
|
| 556 |
+
"loss": 0.6417,
|
| 557 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 558 |
+
"rewards/chosen": -0.37890625,
|
| 559 |
+
"rewards/margins": 0.1455078125,
|
| 560 |
+
"rewards/rejected": -0.5234375,
|
| 561 |
+
"step": 350
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.75,
|
| 565 |
+
"learning_rate": 8.628481651367875e-08,
|
| 566 |
+
"logits/chosen": -1.3828125,
|
| 567 |
+
"logits/rejected": -1.3515625,
|
| 568 |
+
"logps/chosen": -476.0,
|
| 569 |
+
"logps/rejected": -400.0,
|
| 570 |
+
"loss": 0.6398,
|
| 571 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 572 |
+
"rewards/chosen": -0.37890625,
|
| 573 |
+
"rewards/margins": 0.1396484375,
|
| 574 |
+
"rewards/rejected": -0.51953125,
|
| 575 |
+
"step": 360
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.77,
|
| 579 |
+
"learning_rate": 7.289996455765748e-08,
|
| 580 |
+
"logits/chosen": -1.421875,
|
| 581 |
+
"logits/rejected": -1.3984375,
|
| 582 |
+
"logps/chosen": -492.0,
|
| 583 |
+
"logps/rejected": -404.0,
|
| 584 |
+
"loss": 0.6472,
|
| 585 |
+
"rewards/accuracies": 0.625,
|
| 586 |
+
"rewards/chosen": -0.376953125,
|
| 587 |
+
"rewards/margins": 0.138671875,
|
| 588 |
+
"rewards/rejected": -0.515625,
|
| 589 |
+
"step": 370
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 0.8,
|
| 593 |
+
"learning_rate": 6.046442623320145e-08,
|
| 594 |
+
"logits/chosen": -1.4140625,
|
| 595 |
+
"logits/rejected": -1.3984375,
|
| 596 |
+
"logps/chosen": -448.0,
|
| 597 |
+
"logps/rejected": -416.0,
|
| 598 |
+
"loss": 0.6386,
|
| 599 |
+
"rewards/accuracies": 0.640625,
|
| 600 |
+
"rewards/chosen": -0.39453125,
|
| 601 |
+
"rewards/margins": 0.126953125,
|
| 602 |
+
"rewards/rejected": -0.51953125,
|
| 603 |
+
"step": 380
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.82,
|
| 607 |
+
"learning_rate": 4.904486005914027e-08,
|
| 608 |
+
"logits/chosen": -1.3515625,
|
| 609 |
+
"logits/rejected": -1.3515625,
|
| 610 |
+
"logps/chosen": -442.0,
|
| 611 |
+
"logps/rejected": -416.0,
|
| 612 |
+
"loss": 0.6437,
|
| 613 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 614 |
+
"rewards/chosen": -0.39453125,
|
| 615 |
+
"rewards/margins": 0.126953125,
|
| 616 |
+
"rewards/rejected": -0.5234375,
|
| 617 |
+
"step": 390
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.84,
|
| 621 |
+
"learning_rate": 3.8702478614051345e-08,
|
| 622 |
+
"logits/chosen": -1.375,
|
| 623 |
+
"logits/rejected": -1.34375,
|
| 624 |
+
"logps/chosen": -492.0,
|
| 625 |
+
"logps/rejected": -426.0,
|
| 626 |
+
"loss": 0.6462,
|
| 627 |
+
"rewards/accuracies": 0.5625,
|
| 628 |
+
"rewards/chosen": -0.392578125,
|
| 629 |
+
"rewards/margins": 0.1318359375,
|
| 630 |
+
"rewards/rejected": -0.5234375,
|
| 631 |
+
"step": 400
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 0.84,
|
| 635 |
+
"eval_logits/chosen": -1.421875,
|
| 636 |
+
"eval_logits/rejected": -1.375,
|
| 637 |
+
"eval_logps/chosen": -464.0,
|
| 638 |
+
"eval_logps/rejected": -420.0,
|
| 639 |
+
"eval_loss": 0.6495468616485596,
|
| 640 |
+
"eval_rewards/accuracies": 0.6388888955116272,
|
| 641 |
+
"eval_rewards/chosen": -0.400390625,
|
| 642 |
+
"eval_rewards/margins": 0.1552734375,
|
| 643 |
+
"eval_rewards/rejected": -0.5546875,
|
| 644 |
+
"eval_runtime": 51.9691,
|
| 645 |
+
"eval_samples_per_second": 38.484,
|
| 646 |
+
"eval_steps_per_second": 1.212,
|
| 647 |
+
"step": 400
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.86,
|
| 651 |
+
"learning_rate": 2.9492720416985e-08,
|
| 652 |
+
"logits/chosen": -1.328125,
|
| 653 |
+
"logits/rejected": -1.3046875,
|
| 654 |
+
"logps/chosen": -428.0,
|
| 655 |
+
"logps/rejected": -374.0,
|
| 656 |
+
"loss": 0.651,
|
| 657 |
+
"rewards/accuracies": 0.6343749761581421,
|
| 658 |
+
"rewards/chosen": -0.400390625,
|
| 659 |
+
"rewards/margins": 0.130859375,
|
| 660 |
+
"rewards/rejected": -0.53125,
|
| 661 |
+
"step": 410
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.88,
|
| 665 |
+
"learning_rate": 2.1464952759020856e-08,
|
| 666 |
+
"logits/chosen": -1.40625,
|
| 667 |
+
"logits/rejected": -1.375,
|
| 668 |
+
"logps/chosen": -432.0,
|
| 669 |
+
"logps/rejected": -408.0,
|
| 670 |
+
"loss": 0.6437,
|
| 671 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 672 |
+
"rewards/chosen": -0.396484375,
|
| 673 |
+
"rewards/margins": 0.12109375,
|
| 674 |
+
"rewards/rejected": -0.51953125,
|
| 675 |
+
"step": 420
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.9,
|
| 679 |
+
"learning_rate": 1.4662207078575684e-08,
|
| 680 |
+
"logits/chosen": -1.375,
|
| 681 |
+
"logits/rejected": -1.359375,
|
| 682 |
+
"logps/chosen": -458.0,
|
| 683 |
+
"logps/rejected": -404.0,
|
| 684 |
+
"loss": 0.6478,
|
| 685 |
+
"rewards/accuracies": 0.643750011920929,
|
| 686 |
+
"rewards/chosen": -0.380859375,
|
| 687 |
+
"rewards/margins": 0.181640625,
|
| 688 |
+
"rewards/rejected": -0.5625,
|
| 689 |
+
"step": 430
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.92,
|
| 693 |
+
"learning_rate": 9.12094829893642e-09,
|
| 694 |
+
"logits/chosen": -1.421875,
|
| 695 |
+
"logits/rejected": -1.390625,
|
| 696 |
+
"logps/chosen": -420.0,
|
| 697 |
+
"logps/rejected": -392.0,
|
| 698 |
+
"loss": 0.6468,
|
| 699 |
+
"rewards/accuracies": 0.625,
|
| 700 |
+
"rewards/chosen": -0.390625,
|
| 701 |
+
"rewards/margins": 0.138671875,
|
| 702 |
+
"rewards/rejected": -0.52734375,
|
| 703 |
+
"step": 440
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.94,
|
| 707 |
+
"learning_rate": 4.8708793644441086e-09,
|
| 708 |
+
"logits/chosen": -1.4296875,
|
| 709 |
+
"logits/rejected": -1.359375,
|
| 710 |
+
"logps/chosen": -452.0,
|
| 711 |
+
"logps/rejected": -426.0,
|
| 712 |
+
"loss": 0.6587,
|
| 713 |
+
"rewards/accuracies": 0.5843750238418579,
|
| 714 |
+
"rewards/chosen": -0.40234375,
|
| 715 |
+
"rewards/margins": 0.08447265625,
|
| 716 |
+
"rewards/rejected": -0.48828125,
|
| 717 |
+
"step": 450
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.96,
|
| 721 |
+
"learning_rate": 1.9347820230782295e-09,
|
| 722 |
+
"logits/chosen": -1.359375,
|
| 723 |
+
"logits/rejected": -1.3359375,
|
| 724 |
+
"logps/chosen": -426.0,
|
| 725 |
+
"logps/rejected": -390.0,
|
| 726 |
+
"loss": 0.6394,
|
| 727 |
+
"rewards/accuracies": 0.65625,
|
| 728 |
+
"rewards/chosen": -0.40234375,
|
| 729 |
+
"rewards/margins": 0.140625,
|
| 730 |
+
"rewards/rejected": -0.54296875,
|
| 731 |
+
"step": 460
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.98,
|
| 735 |
+
"learning_rate": 3.2839470889836627e-10,
|
| 736 |
+
"logits/chosen": -1.421875,
|
| 737 |
+
"logits/rejected": -1.375,
|
| 738 |
+
"logps/chosen": -470.0,
|
| 739 |
+
"logps/rejected": -422.0,
|
| 740 |
+
"loss": 0.6303,
|
| 741 |
+
"rewards/accuracies": 0.690625011920929,
|
| 742 |
+
"rewards/chosen": -0.36328125,
|
| 743 |
+
"rewards/margins": 0.19921875,
|
| 744 |
+
"rewards/rejected": -0.55859375,
|
| 745 |
+
"step": 470
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 1.0,
|
| 749 |
+
"step": 477,
|
| 750 |
+
"total_flos": 0.0,
|
| 751 |
+
"train_loss": 0.6625334221861897,
|
| 752 |
+
"train_runtime": 3684.7873,
|
| 753 |
+
"train_samples_per_second": 16.591,
|
| 754 |
+
"train_steps_per_second": 0.129
|
| 755 |
+
}
|
| 756 |
+
],
|
| 757 |
+
"logging_steps": 10,
|
| 758 |
+
"max_steps": 477,
|
| 759 |
+
"num_input_tokens_seen": 0,
|
| 760 |
+
"num_train_epochs": 1,
|
| 761 |
+
"save_steps": 100,
|
| 762 |
+
"total_flos": 0.0,
|
| 763 |
+
"train_batch_size": 8,
|
| 764 |
+
"trial_name": null,
|
| 765 |
+
"trial_params": null
|
| 766 |
+
}
|