Commit
•
7deebed
1
Parent(s):
eb7443d
Model save
Browse files
README.md
ADDED
|
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: peft
|
| 3 |
+
base_model: HuggingFaceTB/smollm2-1.7B-8k-mix7-ep2-v2
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
model-index:
|
| 9 |
+
- name: smollm2-1.7B-8k-mix7-ep2-v2-qlora-r16-a16-lr3e4-mix1-dpo
|
| 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 |
+
# smollm2-1.7B-8k-mix7-ep2-v2-qlora-r16-a16-lr3e4-mix1-dpo
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [HuggingFaceTB/smollm2-1.7B-8k-mix7-ep2-v2](https://huggingface.co/HuggingFaceTB/smollm2-1.7B-8k-mix7-ep2-v2) on an unknown dataset.
|
| 19 |
+
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.5157
|
| 21 |
+
- Rewards/chosen: -4.3456
|
| 22 |
+
- Rewards/rejected: -5.3593
|
| 23 |
+
- Rewards/accuracies: 0.7695
|
| 24 |
+
- Rewards/margins: 1.0137
|
| 25 |
+
- Logps/rejected: -842.1772
|
| 26 |
+
- Logps/chosen: -747.2964
|
| 27 |
+
- Logits/rejected: -1.2109
|
| 28 |
+
- Logits/chosen: -1.1871
|
| 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: 0.0003
|
| 48 |
+
- train_batch_size: 4
|
| 49 |
+
- eval_batch_size: 8
|
| 50 |
+
- seed: 42
|
| 51 |
+
- distributed_type: multi-GPU
|
| 52 |
+
- num_devices: 8
|
| 53 |
+
- gradient_accumulation_steps: 4
|
| 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 |
+
| 0.5629 | 0.2093 | 100 | 0.5765 | -2.2363 | -2.7515 | 0.7227 | 0.5152 | -581.3958 | -536.3666 | -1.6836 | -1.6572 |
|
| 66 |
+
| 0.5499 | 0.4186 | 200 | 0.5310 | -3.6681 | -4.4704 | 0.7734 | 0.8022 | -753.2846 | -679.5535 | -0.8209 | -0.8334 |
|
| 67 |
+
| 0.511 | 0.6279 | 300 | 0.5225 | -4.8153 | -5.7685 | 0.7695 | 0.9532 | -883.0991 | -794.2732 | -1.1394 | -1.1157 |
|
| 68 |
+
| 0.508 | 0.8373 | 400 | 0.5157 | -4.3456 | -5.3593 | 0.7695 | 1.0137 | -842.1772 | -747.2964 | -1.2109 | -1.1871 |
|
| 69 |
+
|
| 70 |
+
|
| 71 |
+
### Framework versions
|
| 72 |
+
|
| 73 |
+
- PEFT 0.13.2
|
| 74 |
+
- Transformers 4.45.0
|
| 75 |
+
- Pytorch 2.4.1+cu121
|
| 76 |
+
- Datasets 3.0.2
|
| 77 |
+
- Tokenizers 0.20.1
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 72396376
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6db4af8d12501a8693f3206a9f20b5fb2aa6c5f87d058a88925c0bead39f39d6
|
| 3 |
size 72396376
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9984301412872841,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.5423707397229016,
|
| 5 |
+
"train_runtime": 2137.6738,
|
| 6 |
+
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 28.598,
|
| 8 |
+
"train_steps_per_second": 0.223
|
| 9 |
+
}
|
runs/Nov05_11-27-51_ip-26-0-167-175/events.out.tfevents.1730806544.ip-26-0-167-175.3215185.0
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:f6fdf40fbc0e92af70ac9ea50d7af522f058307daf81936b604ce8b09908a442
|
| 3 |
+
size 42734
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9984301412872841,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.5423707397229016,
|
| 5 |
+
"train_runtime": 2137.6738,
|
| 6 |
+
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 28.598,
|
| 8 |
+
"train_steps_per_second": 0.223
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,826 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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": 0.0020931449502878076,
|
| 13 |
+
"grad_norm": 0.035678211599588394,
|
| 14 |
+
"learning_rate": 6.2499999999999995e-06,
|
| 15 |
+
"logits/chosen": -1.1021082401275635,
|
| 16 |
+
"logits/rejected": -0.9669773578643799,
|
| 17 |
+
"logps/chosen": -392.66217041015625,
|
| 18 |
+
"logps/rejected": -388.5433654785156,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.020931449502878074,
|
| 28 |
+
"grad_norm": 0.03580914065241814,
|
| 29 |
+
"learning_rate": 6.25e-05,
|
| 30 |
+
"logits/chosen": -1.0458121299743652,
|
| 31 |
+
"logits/rejected": -1.1097700595855713,
|
| 32 |
+
"logps/chosen": -340.9582214355469,
|
| 33 |
+
"logps/rejected": -299.7181396484375,
|
| 34 |
+
"loss": 0.6929,
|
| 35 |
+
"rewards/accuracies": 0.4444444477558136,
|
| 36 |
+
"rewards/chosen": 0.0002490293118171394,
|
| 37 |
+
"rewards/margins": 0.0008229945669882,
|
| 38 |
+
"rewards/rejected": -0.0005739651969633996,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.04186289900575615,
|
| 43 |
+
"grad_norm": 0.037438321858644485,
|
| 44 |
+
"learning_rate": 0.000125,
|
| 45 |
+
"logits/chosen": -0.9009532928466797,
|
| 46 |
+
"logits/rejected": -1.0806249380111694,
|
| 47 |
+
"logps/chosen": -355.1332092285156,
|
| 48 |
+
"logps/rejected": -306.08709716796875,
|
| 49 |
+
"loss": 0.6915,
|
| 50 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 51 |
+
"rewards/chosen": -0.0018794331699609756,
|
| 52 |
+
"rewards/margins": 0.002060088561847806,
|
| 53 |
+
"rewards/rejected": -0.003939521964639425,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.06279434850863422,
|
| 58 |
+
"grad_norm": 0.05021581053733826,
|
| 59 |
+
"learning_rate": 0.00018749999999999998,
|
| 60 |
+
"logits/chosen": -0.9453525543212891,
|
| 61 |
+
"logits/rejected": -1.0644036531448364,
|
| 62 |
+
"logps/chosen": -315.98760986328125,
|
| 63 |
+
"logps/rejected": -268.6424865722656,
|
| 64 |
+
"loss": 0.6852,
|
| 65 |
+
"rewards/accuracies": 0.581250011920929,
|
| 66 |
+
"rewards/chosen": -0.01840055361390114,
|
| 67 |
+
"rewards/margins": 0.018759142607450485,
|
| 68 |
+
"rewards/rejected": -0.03715970367193222,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.0837257980115123,
|
| 73 |
+
"grad_norm": 0.08909210562705994,
|
| 74 |
+
"learning_rate": 0.00025,
|
| 75 |
+
"logits/chosen": -1.0098339319229126,
|
| 76 |
+
"logits/rejected": -1.0776870250701904,
|
| 77 |
+
"logps/chosen": -317.6771240234375,
|
| 78 |
+
"logps/rejected": -308.1802673339844,
|
| 79 |
+
"loss": 0.6584,
|
| 80 |
+
"rewards/accuracies": 0.625,
|
| 81 |
+
"rewards/chosen": -0.16074401140213013,
|
| 82 |
+
"rewards/margins": 0.09125810116529465,
|
| 83 |
+
"rewards/rejected": -0.25200212001800537,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.10465724751439037,
|
| 88 |
+
"grad_norm": 0.1175214871764183,
|
| 89 |
+
"learning_rate": 0.0002999839121261415,
|
| 90 |
+
"logits/chosen": -1.0516436100006104,
|
| 91 |
+
"logits/rejected": -1.1754628419876099,
|
| 92 |
+
"logps/chosen": -393.256103515625,
|
| 93 |
+
"logps/rejected": -375.66510009765625,
|
| 94 |
+
"loss": 0.6346,
|
| 95 |
+
"rewards/accuracies": 0.706250011920929,
|
| 96 |
+
"rewards/chosen": -0.458698570728302,
|
| 97 |
+
"rewards/margins": 0.20651093125343323,
|
| 98 |
+
"rewards/rejected": -0.6652094125747681,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.12558869701726844,
|
| 103 |
+
"grad_norm": 0.1383933126926422,
|
| 104 |
+
"learning_rate": 0.00029942119880575817,
|
| 105 |
+
"logits/chosen": -1.0281826257705688,
|
| 106 |
+
"logits/rejected": -1.0935122966766357,
|
| 107 |
+
"logps/chosen": -332.78228759765625,
|
| 108 |
+
"logps/rejected": -345.4633483886719,
|
| 109 |
+
"loss": 0.6227,
|
| 110 |
+
"rewards/accuracies": 0.75,
|
| 111 |
+
"rewards/chosen": -0.49291595816612244,
|
| 112 |
+
"rewards/margins": 0.23161530494689941,
|
| 113 |
+
"rewards/rejected": -0.7245312929153442,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.14652014652014653,
|
| 118 |
+
"grad_norm": 0.1275259107351303,
|
| 119 |
+
"learning_rate": 0.00029805753939568687,
|
| 120 |
+
"logits/chosen": -1.4444881677627563,
|
| 121 |
+
"logits/rejected": -1.4646410942077637,
|
| 122 |
+
"logps/chosen": -412.0874938964844,
|
| 123 |
+
"logps/rejected": -426.487548828125,
|
| 124 |
+
"loss": 0.6048,
|
| 125 |
+
"rewards/accuracies": 0.668749988079071,
|
| 126 |
+
"rewards/chosen": -0.7971259951591492,
|
| 127 |
+
"rewards/margins": 0.35752782225608826,
|
| 128 |
+
"rewards/rejected": -1.154653787612915,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.1674515960230246,
|
| 133 |
+
"grad_norm": 0.17714321613311768,
|
| 134 |
+
"learning_rate": 0.00029590024355266256,
|
| 135 |
+
"logits/chosen": -1.1742874383926392,
|
| 136 |
+
"logits/rejected": -1.4253867864608765,
|
| 137 |
+
"logps/chosen": -479.925537109375,
|
| 138 |
+
"logps/rejected": -425.94598388671875,
|
| 139 |
+
"loss": 0.583,
|
| 140 |
+
"rewards/accuracies": 0.731249988079071,
|
| 141 |
+
"rewards/chosen": -1.072460412979126,
|
| 142 |
+
"rewards/margins": 0.462944358587265,
|
| 143 |
+
"rewards/rejected": -1.5354046821594238,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.18838304552590268,
|
| 148 |
+
"grad_norm": 0.2510909140110016,
|
| 149 |
+
"learning_rate": 0.0002929608750821129,
|
| 150 |
+
"logits/chosen": -1.7680635452270508,
|
| 151 |
+
"logits/rejected": -1.7849355936050415,
|
| 152 |
+
"logps/chosen": -501.5375061035156,
|
| 153 |
+
"logps/rejected": -569.551513671875,
|
| 154 |
+
"loss": 0.5604,
|
| 155 |
+
"rewards/accuracies": 0.706250011920929,
|
| 156 |
+
"rewards/chosen": -2.1197433471679688,
|
| 157 |
+
"rewards/margins": 0.7176006436347961,
|
| 158 |
+
"rewards/rejected": -2.83734393119812,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.20931449502878074,
|
| 163 |
+
"grad_norm": 0.1974615901708603,
|
| 164 |
+
"learning_rate": 0.0002892551899524109,
|
| 165 |
+
"logits/chosen": -1.4390876293182373,
|
| 166 |
+
"logits/rejected": -1.5163460969924927,
|
| 167 |
+
"logps/chosen": -541.5751953125,
|
| 168 |
+
"logps/rejected": -558.0751953125,
|
| 169 |
+
"loss": 0.5629,
|
| 170 |
+
"rewards/accuracies": 0.762499988079071,
|
| 171 |
+
"rewards/chosen": -1.9338293075561523,
|
| 172 |
+
"rewards/margins": 0.575296938419342,
|
| 173 |
+
"rewards/rejected": -2.5091261863708496,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.20931449502878074,
|
| 178 |
+
"eval_logits/chosen": -1.6571975946426392,
|
| 179 |
+
"eval_logits/rejected": -1.6835596561431885,
|
| 180 |
+
"eval_logps/chosen": -536.3665771484375,
|
| 181 |
+
"eval_logps/rejected": -581.395751953125,
|
| 182 |
+
"eval_loss": 0.5765172243118286,
|
| 183 |
+
"eval_rewards/accuracies": 0.72265625,
|
| 184 |
+
"eval_rewards/chosen": -2.2362735271453857,
|
| 185 |
+
"eval_rewards/margins": 0.5152010917663574,
|
| 186 |
+
"eval_rewards/rejected": -2.751474618911743,
|
| 187 |
+
"eval_runtime": 35.9544,
|
| 188 |
+
"eval_samples_per_second": 55.626,
|
| 189 |
+
"eval_steps_per_second": 0.89,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.2302459445316588,
|
| 194 |
+
"grad_norm": 0.20197245478630066,
|
| 195 |
+
"learning_rate": 0.00028480305183777385,
|
| 196 |
+
"logits/chosen": -1.446223497390747,
|
| 197 |
+
"logits/rejected": -1.5271519422531128,
|
| 198 |
+
"logps/chosen": -602.8939208984375,
|
| 199 |
+
"logps/rejected": -629.87158203125,
|
| 200 |
+
"loss": 0.5644,
|
| 201 |
+
"rewards/accuracies": 0.706250011920929,
|
| 202 |
+
"rewards/chosen": -2.7250099182128906,
|
| 203 |
+
"rewards/margins": 0.5918947458267212,
|
| 204 |
+
"rewards/rejected": -3.3169047832489014,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.25117739403453687,
|
| 209 |
+
"grad_norm": 0.2141677737236023,
|
| 210 |
+
"learning_rate": 0.00027962832564252724,
|
| 211 |
+
"logits/chosen": -1.6662397384643555,
|
| 212 |
+
"logits/rejected": -1.6905275583267212,
|
| 213 |
+
"logps/chosen": -632.6597900390625,
|
| 214 |
+
"logps/rejected": -688.1976318359375,
|
| 215 |
+
"loss": 0.5477,
|
| 216 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 217 |
+
"rewards/chosen": -3.3453495502471924,
|
| 218 |
+
"rewards/margins": 0.6758114695549011,
|
| 219 |
+
"rewards/rejected": -4.021161079406738,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.272108843537415,
|
| 224 |
+
"grad_norm": 0.20801568031311035,
|
| 225 |
+
"learning_rate": 0.0002737587495774764,
|
| 226 |
+
"logits/chosen": -1.4505422115325928,
|
| 227 |
+
"logits/rejected": -1.4989721775054932,
|
| 228 |
+
"logps/chosen": -658.4010009765625,
|
| 229 |
+
"logps/rejected": -700.3973388671875,
|
| 230 |
+
"loss": 0.5112,
|
| 231 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 232 |
+
"rewards/chosen": -3.428065776824951,
|
| 233 |
+
"rewards/margins": 0.6822351217269897,
|
| 234 |
+
"rewards/rejected": -4.1103010177612305,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.29304029304029305,
|
| 239 |
+
"grad_norm": 0.24513058364391327,
|
| 240 |
+
"learning_rate": 0.0002672257864741005,
|
| 241 |
+
"logits/chosen": -1.4723621606826782,
|
| 242 |
+
"logits/rejected": -1.5368088483810425,
|
| 243 |
+
"logps/chosen": -734.0245361328125,
|
| 244 |
+
"logps/rejected": -741.5623779296875,
|
| 245 |
+
"loss": 0.5449,
|
| 246 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 247 |
+
"rewards/chosen": -3.7274997234344482,
|
| 248 |
+
"rewards/margins": 0.7957648634910583,
|
| 249 |
+
"rewards/rejected": -4.5232648849487305,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.3139717425431711,
|
| 254 |
+
"grad_norm": 0.2503284513950348,
|
| 255 |
+
"learning_rate": 0.0002600644551335706,
|
| 256 |
+
"logits/chosen": -1.2388180494308472,
|
| 257 |
+
"logits/rejected": -1.3000439405441284,
|
| 258 |
+
"logps/chosen": -629.1339111328125,
|
| 259 |
+
"logps/rejected": -636.8096923828125,
|
| 260 |
+
"loss": 0.5479,
|
| 261 |
+
"rewards/accuracies": 0.699999988079071,
|
| 262 |
+
"rewards/chosen": -3.0619120597839355,
|
| 263 |
+
"rewards/margins": 0.5719773173332214,
|
| 264 |
+
"rewards/rejected": -3.6338894367218018,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.3349031920460492,
|
| 269 |
+
"grad_norm": 0.18707025051116943,
|
| 270 |
+
"learning_rate": 0.0002523131426146173,
|
| 271 |
+
"logits/chosen": -1.1968213319778442,
|
| 272 |
+
"logits/rejected": -1.3508607149124146,
|
| 273 |
+
"logps/chosen": -642.1495971679688,
|
| 274 |
+
"logps/rejected": -680.3194580078125,
|
| 275 |
+
"loss": 0.5573,
|
| 276 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 277 |
+
"rewards/chosen": -3.2288098335266113,
|
| 278 |
+
"rewards/margins": 0.7885423898696899,
|
| 279 |
+
"rewards/rejected": -4.017352104187012,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.35583464154892724,
|
| 284 |
+
"grad_norm": 0.18305683135986328,
|
| 285 |
+
"learning_rate": 0.00024401339846644537,
|
| 286 |
+
"logits/chosen": -1.2790342569351196,
|
| 287 |
+
"logits/rejected": -1.3211687803268433,
|
| 288 |
+
"logps/chosen": -597.0249633789062,
|
| 289 |
+
"logps/rejected": -608.4508666992188,
|
| 290 |
+
"loss": 0.5409,
|
| 291 |
+
"rewards/accuracies": 0.706250011920929,
|
| 292 |
+
"rewards/chosen": -3.0255608558654785,
|
| 293 |
+
"rewards/margins": 0.49673914909362793,
|
| 294 |
+
"rewards/rejected": -3.5222995281219482,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.37676609105180536,
|
| 299 |
+
"grad_norm": 0.2515067458152771,
|
| 300 |
+
"learning_rate": 0.00023520971200967334,
|
| 301 |
+
"logits/chosen": -1.1840099096298218,
|
| 302 |
+
"logits/rejected": -1.2452869415283203,
|
| 303 |
+
"logps/chosen": -684.8968505859375,
|
| 304 |
+
"logps/rejected": -719.9307861328125,
|
| 305 |
+
"loss": 0.5217,
|
| 306 |
+
"rewards/accuracies": 0.75,
|
| 307 |
+
"rewards/chosen": -3.799819231033325,
|
| 308 |
+
"rewards/margins": 0.6900444030761719,
|
| 309 |
+
"rewards/rejected": -4.489863872528076,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.3976975405546834,
|
| 314 |
+
"grad_norm": 0.21723517775535583,
|
| 315 |
+
"learning_rate": 0.00022594927385914543,
|
| 316 |
+
"logits/chosen": -1.2460777759552002,
|
| 317 |
+
"logits/rejected": -1.2459148168563843,
|
| 318 |
+
"logps/chosen": -649.2003784179688,
|
| 319 |
+
"logps/rejected": -708.8577880859375,
|
| 320 |
+
"loss": 0.5298,
|
| 321 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 322 |
+
"rewards/chosen": -3.549588441848755,
|
| 323 |
+
"rewards/margins": 0.6792875528335571,
|
| 324 |
+
"rewards/rejected": -4.22887659072876,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.4186289900575615,
|
| 329 |
+
"grad_norm": 0.26195383071899414,
|
| 330 |
+
"learning_rate": 0.0002162817229669295,
|
| 331 |
+
"logits/chosen": -0.8848621249198914,
|
| 332 |
+
"logits/rejected": -0.8319551348686218,
|
| 333 |
+
"logps/chosen": -649.4385986328125,
|
| 334 |
+
"logps/rejected": -711.3386840820312,
|
| 335 |
+
"loss": 0.5499,
|
| 336 |
+
"rewards/accuracies": 0.65625,
|
| 337 |
+
"rewards/chosen": -3.6400089263916016,
|
| 338 |
+
"rewards/margins": 0.5325465798377991,
|
| 339 |
+
"rewards/rejected": -4.172555923461914,
|
| 340 |
+
"step": 200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.4186289900575615,
|
| 344 |
+
"eval_logits/chosen": -0.8333719968795776,
|
| 345 |
+
"eval_logits/rejected": -0.8208783268928528,
|
| 346 |
+
"eval_logps/chosen": -679.553466796875,
|
| 347 |
+
"eval_logps/rejected": -753.2846069335938,
|
| 348 |
+
"eval_loss": 0.5309607982635498,
|
| 349 |
+
"eval_rewards/accuracies": 0.7734375,
|
| 350 |
+
"eval_rewards/chosen": -3.6681408882141113,
|
| 351 |
+
"eval_rewards/margins": 0.8022220730781555,
|
| 352 |
+
"eval_rewards/rejected": -4.470362663269043,
|
| 353 |
+
"eval_runtime": 34.0707,
|
| 354 |
+
"eval_samples_per_second": 58.702,
|
| 355 |
+
"eval_steps_per_second": 0.939,
|
| 356 |
+
"step": 200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.43956043956043955,
|
| 360 |
+
"grad_norm": 0.19068197906017303,
|
| 361 |
+
"learning_rate": 0.00020625888054143427,
|
| 362 |
+
"logits/chosen": -0.8465880155563354,
|
| 363 |
+
"logits/rejected": -0.9894304275512695,
|
| 364 |
+
"logps/chosen": -744.5973510742188,
|
| 365 |
+
"logps/rejected": -742.8700561523438,
|
| 366 |
+
"loss": 0.5433,
|
| 367 |
+
"rewards/accuracies": 0.737500011920929,
|
| 368 |
+
"rewards/chosen": -3.855426788330078,
|
| 369 |
+
"rewards/margins": 0.6753937005996704,
|
| 370 |
+
"rewards/rejected": -4.530820369720459,
|
| 371 |
+
"step": 210
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.4604918890633176,
|
| 375 |
+
"grad_norm": 0.22149965167045593,
|
| 376 |
+
"learning_rate": 0.00019593447226892385,
|
| 377 |
+
"logits/chosen": -0.9477745294570923,
|
| 378 |
+
"logits/rejected": -0.9140194058418274,
|
| 379 |
+
"logps/chosen": -622.2196044921875,
|
| 380 |
+
"logps/rejected": -685.7794189453125,
|
| 381 |
+
"loss": 0.535,
|
| 382 |
+
"rewards/accuracies": 0.737500011920929,
|
| 383 |
+
"rewards/chosen": -3.5976402759552,
|
| 384 |
+
"rewards/margins": 0.6984301805496216,
|
| 385 |
+
"rewards/rejected": -4.296070575714111,
|
| 386 |
+
"step": 220
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.48142333856619574,
|
| 390 |
+
"grad_norm": 0.26226335763931274,
|
| 391 |
+
"learning_rate": 0.0001853638403264141,
|
| 392 |
+
"logits/chosen": -0.8337217569351196,
|
| 393 |
+
"logits/rejected": -0.8363305330276489,
|
| 394 |
+
"logps/chosen": -682.59423828125,
|
| 395 |
+
"logps/rejected": -737.8418579101562,
|
| 396 |
+
"loss": 0.5457,
|
| 397 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 398 |
+
"rewards/chosen": -3.968369722366333,
|
| 399 |
+
"rewards/margins": 0.8683279156684875,
|
| 400 |
+
"rewards/rejected": -4.836697578430176,
|
| 401 |
+
"step": 230
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.5023547880690737,
|
| 405 |
+
"grad_norm": 0.21782232820987701,
|
| 406 |
+
"learning_rate": 0.00017460364672965327,
|
| 407 |
+
"logits/chosen": -0.799470841884613,
|
| 408 |
+
"logits/rejected": -0.8606585264205933,
|
| 409 |
+
"logps/chosen": -692.9552001953125,
|
| 410 |
+
"logps/rejected": -750.71484375,
|
| 411 |
+
"loss": 0.5022,
|
| 412 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 413 |
+
"rewards/chosen": -3.8266777992248535,
|
| 414 |
+
"rewards/margins": 0.792169451713562,
|
| 415 |
+
"rewards/rejected": -4.618846893310547,
|
| 416 |
+
"step": 240
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.5232862375719518,
|
| 420 |
+
"grad_norm": 0.2562369406223297,
|
| 421 |
+
"learning_rate": 0.00016371156960634018,
|
| 422 |
+
"logits/chosen": -1.0665327310562134,
|
| 423 |
+
"logits/rejected": -1.1944453716278076,
|
| 424 |
+
"logps/chosen": -768.0906982421875,
|
| 425 |
+
"logps/rejected": -832.09521484375,
|
| 426 |
+
"loss": 0.5107,
|
| 427 |
+
"rewards/accuracies": 0.8125,
|
| 428 |
+
"rewards/chosen": -4.486889839172363,
|
| 429 |
+
"rewards/margins": 1.1490213871002197,
|
| 430 |
+
"rewards/rejected": -5.635911464691162,
|
| 431 |
+
"step": 250
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.54421768707483,
|
| 435 |
+
"grad_norm": 0.29225146770477295,
|
| 436 |
+
"learning_rate": 0.0001527459940226516,
|
| 437 |
+
"logits/chosen": -1.1020123958587646,
|
| 438 |
+
"logits/rejected": -1.1401180028915405,
|
| 439 |
+
"logps/chosen": -767.971435546875,
|
| 440 |
+
"logps/rejected": -831.6994018554688,
|
| 441 |
+
"loss": 0.5209,
|
| 442 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 443 |
+
"rewards/chosen": -4.771060943603516,
|
| 444 |
+
"rewards/margins": 0.8642056584358215,
|
| 445 |
+
"rewards/rejected": -5.635266304016113,
|
| 446 |
+
"step": 260
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.565149136577708,
|
| 450 |
+
"grad_norm": 0.1788558065891266,
|
| 451 |
+
"learning_rate": 0.00014176569902035088,
|
| 452 |
+
"logits/chosen": -1.176230549812317,
|
| 453 |
+
"logits/rejected": -1.214536428451538,
|
| 454 |
+
"logps/chosen": -739.462646484375,
|
| 455 |
+
"logps/rejected": -794.5394287109375,
|
| 456 |
+
"loss": 0.5172,
|
| 457 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 458 |
+
"rewards/chosen": -4.040095329284668,
|
| 459 |
+
"rewards/margins": 0.9585875272750854,
|
| 460 |
+
"rewards/rejected": -4.998682975769043,
|
| 461 |
+
"step": 270
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.5860805860805861,
|
| 465 |
+
"grad_norm": 0.22155936062335968,
|
| 466 |
+
"learning_rate": 0.0001308295425420593,
|
| 467 |
+
"logits/chosen": -1.1204885244369507,
|
| 468 |
+
"logits/rejected": -1.2108111381530762,
|
| 469 |
+
"logps/chosen": -661.9423217773438,
|
| 470 |
+
"logps/rejected": -689.7301635742188,
|
| 471 |
+
"loss": 0.5431,
|
| 472 |
+
"rewards/accuracies": 0.668749988079071,
|
| 473 |
+
"rewards/chosen": -3.6794509887695312,
|
| 474 |
+
"rewards/margins": 0.6488171219825745,
|
| 475 |
+
"rewards/rejected": -4.328268527984619,
|
| 476 |
+
"step": 280
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.6070120355834642,
|
| 480 |
+
"grad_norm": 0.2569100558757782,
|
| 481 |
+
"learning_rate": 0.00011999614593359334,
|
| 482 |
+
"logits/chosen": -1.0885214805603027,
|
| 483 |
+
"logits/rejected": -1.1296148300170898,
|
| 484 |
+
"logps/chosen": -734.2669067382812,
|
| 485 |
+
"logps/rejected": -823.91015625,
|
| 486 |
+
"loss": 0.5135,
|
| 487 |
+
"rewards/accuracies": 0.768750011920929,
|
| 488 |
+
"rewards/chosen": -4.625601291656494,
|
| 489 |
+
"rewards/margins": 0.7508729100227356,
|
| 490 |
+
"rewards/rejected": -5.376473903656006,
|
| 491 |
+
"step": 290
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.6279434850863422,
|
| 495 |
+
"grad_norm": 0.20601074397563934,
|
| 496 |
+
"learning_rate": 0.00010932357971453744,
|
| 497 |
+
"logits/chosen": -1.036828875541687,
|
| 498 |
+
"logits/rejected": -1.1336538791656494,
|
| 499 |
+
"logps/chosen": -893.9344482421875,
|
| 500 |
+
"logps/rejected": -916.6799926757812,
|
| 501 |
+
"loss": 0.511,
|
| 502 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 503 |
+
"rewards/chosen": -5.284546852111816,
|
| 504 |
+
"rewards/margins": 0.8582801818847656,
|
| 505 |
+
"rewards/rejected": -6.142827033996582,
|
| 506 |
+
"step": 300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.6279434850863422,
|
| 510 |
+
"eval_logits/chosen": -1.115662693977356,
|
| 511 |
+
"eval_logits/rejected": -1.1394394636154175,
|
| 512 |
+
"eval_logps/chosen": -794.273193359375,
|
| 513 |
+
"eval_logps/rejected": -883.09912109375,
|
| 514 |
+
"eval_loss": 0.5225423574447632,
|
| 515 |
+
"eval_rewards/accuracies": 0.76953125,
|
| 516 |
+
"eval_rewards/chosen": -4.815339088439941,
|
| 517 |
+
"eval_rewards/margins": 0.953169584274292,
|
| 518 |
+
"eval_rewards/rejected": -5.768507957458496,
|
| 519 |
+
"eval_runtime": 34.0232,
|
| 520 |
+
"eval_samples_per_second": 58.783,
|
| 521 |
+
"eval_steps_per_second": 0.941,
|
| 522 |
+
"step": 300
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.6488749345892203,
|
| 526 |
+
"grad_norm": 0.2347152978181839,
|
| 527 |
+
"learning_rate": 9.886905230142432e-05,
|
| 528 |
+
"logits/chosen": -1.124932050704956,
|
| 529 |
+
"logits/rejected": -1.1792018413543701,
|
| 530 |
+
"logps/chosen": -800.5477905273438,
|
| 531 |
+
"logps/rejected": -829.0084228515625,
|
| 532 |
+
"loss": 0.5016,
|
| 533 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 534 |
+
"rewards/chosen": -4.595987796783447,
|
| 535 |
+
"rewards/margins": 0.9579511880874634,
|
| 536 |
+
"rewards/rejected": -5.553939342498779,
|
| 537 |
+
"step": 310
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.6698063840920984,
|
| 541 |
+
"grad_norm": 0.22556880116462708,
|
| 542 |
+
"learning_rate": 8.868860335206678e-05,
|
| 543 |
+
"logits/chosen": -0.987607479095459,
|
| 544 |
+
"logits/rejected": -1.0762406587600708,
|
| 545 |
+
"logps/chosen": -740.9945068359375,
|
| 546 |
+
"logps/rejected": -809.6924438476562,
|
| 547 |
+
"loss": 0.49,
|
| 548 |
+
"rewards/accuracies": 0.78125,
|
| 549 |
+
"rewards/chosen": -4.5622124671936035,
|
| 550 |
+
"rewards/margins": 1.0737111568450928,
|
| 551 |
+
"rewards/rejected": -5.635923385620117,
|
| 552 |
+
"step": 320
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.6907378335949764,
|
| 556 |
+
"grad_norm": 0.22182781994342804,
|
| 557 |
+
"learning_rate": 7.883680337481599e-05,
|
| 558 |
+
"logits/chosen": -1.090471863746643,
|
| 559 |
+
"logits/rejected": -1.1103293895721436,
|
| 560 |
+
"logps/chosen": -724.054931640625,
|
| 561 |
+
"logps/rejected": -794.3676147460938,
|
| 562 |
+
"loss": 0.5071,
|
| 563 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 564 |
+
"rewards/chosen": -4.166729927062988,
|
| 565 |
+
"rewards/margins": 0.8365222215652466,
|
| 566 |
+
"rewards/rejected": -5.003252029418945,
|
| 567 |
+
"step": 330
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.7116692830978545,
|
| 571 |
+
"grad_norm": 0.280301034450531,
|
| 572 |
+
"learning_rate": 6.936646121293653e-05,
|
| 573 |
+
"logits/chosen": -1.1425997018814087,
|
| 574 |
+
"logits/rejected": -1.207086205482483,
|
| 575 |
+
"logps/chosen": -778.4783935546875,
|
| 576 |
+
"logps/rejected": -811.2172241210938,
|
| 577 |
+
"loss": 0.5069,
|
| 578 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 579 |
+
"rewards/chosen": -4.307219505310059,
|
| 580 |
+
"rewards/margins": 0.8611894845962524,
|
| 581 |
+
"rewards/rejected": -5.168410301208496,
|
| 582 |
+
"step": 340
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 0.7326007326007326,
|
| 586 |
+
"grad_norm": 0.2707894444465637,
|
| 587 |
+
"learning_rate": 6.032834097207889e-05,
|
| 588 |
+
"logits/chosen": -1.2494275569915771,
|
| 589 |
+
"logits/rejected": -1.364678144454956,
|
| 590 |
+
"logps/chosen": -731.4861450195312,
|
| 591 |
+
"logps/rejected": -781.0096435546875,
|
| 592 |
+
"loss": 0.4841,
|
| 593 |
+
"rewards/accuracies": 0.793749988079071,
|
| 594 |
+
"rewards/chosen": -4.36992073059082,
|
| 595 |
+
"rewards/margins": 0.9430965185165405,
|
| 596 |
+
"rewards/rejected": -5.313016891479492,
|
| 597 |
+
"step": 350
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.7535321821036107,
|
| 601 |
+
"grad_norm": 0.17782525718212128,
|
| 602 |
+
"learning_rate": 5.1770889908207245e-05,
|
| 603 |
+
"logits/chosen": -1.1481292247772217,
|
| 604 |
+
"logits/rejected": -1.296864628791809,
|
| 605 |
+
"logps/chosen": -804.0589599609375,
|
| 606 |
+
"logps/rejected": -836.4563598632812,
|
| 607 |
+
"loss": 0.5281,
|
| 608 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 609 |
+
"rewards/chosen": -4.434665203094482,
|
| 610 |
+
"rewards/margins": 0.9478410482406616,
|
| 611 |
+
"rewards/rejected": -5.382506370544434,
|
| 612 |
+
"step": 360
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.7744636316064888,
|
| 616 |
+
"grad_norm": 0.19104325771331787,
|
| 617 |
+
"learning_rate": 4.373997873459449e-05,
|
| 618 |
+
"logits/chosen": -1.1908161640167236,
|
| 619 |
+
"logits/rejected": -1.3131165504455566,
|
| 620 |
+
"logps/chosen": -732.2794189453125,
|
| 621 |
+
"logps/rejected": -775.1005859375,
|
| 622 |
+
"loss": 0.5126,
|
| 623 |
+
"rewards/accuracies": 0.75,
|
| 624 |
+
"rewards/chosen": -4.367339134216309,
|
| 625 |
+
"rewards/margins": 0.8501862287521362,
|
| 626 |
+
"rewards/rejected": -5.217526435852051,
|
| 627 |
+
"step": 370
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.7953950811093669,
|
| 631 |
+
"grad_norm": 0.2322331815958023,
|
| 632 |
+
"learning_rate": 3.6278655739920866e-05,
|
| 633 |
+
"logits/chosen": -1.2175922393798828,
|
| 634 |
+
"logits/rejected": -1.2287733554840088,
|
| 635 |
+
"logps/chosen": -763.4591674804688,
|
| 636 |
+
"logps/rejected": -873.5496215820312,
|
| 637 |
+
"loss": 0.5018,
|
| 638 |
+
"rewards/accuracies": 0.699999988079071,
|
| 639 |
+
"rewards/chosen": -4.82603120803833,
|
| 640 |
+
"rewards/margins": 0.9868711233139038,
|
| 641 |
+
"rewards/rejected": -5.812902927398682,
|
| 642 |
+
"step": 380
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.8163265306122449,
|
| 646 |
+
"grad_norm": 0.2882665693759918,
|
| 647 |
+
"learning_rate": 2.942691603548416e-05,
|
| 648 |
+
"logits/chosen": -1.0723416805267334,
|
| 649 |
+
"logits/rejected": -1.1688863039016724,
|
| 650 |
+
"logps/chosen": -854.5564575195312,
|
| 651 |
+
"logps/rejected": -905.0885009765625,
|
| 652 |
+
"loss": 0.5087,
|
| 653 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 654 |
+
"rewards/chosen": -4.80784273147583,
|
| 655 |
+
"rewards/margins": 0.8358656167984009,
|
| 656 |
+
"rewards/rejected": -5.643708229064941,
|
| 657 |
+
"step": 390
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.837257980115123,
|
| 661 |
+
"grad_norm": 0.2635945677757263,
|
| 662 |
+
"learning_rate": 2.3221487168430806e-05,
|
| 663 |
+
"logits/chosen": -1.185377836227417,
|
| 664 |
+
"logits/rejected": -1.252471685409546,
|
| 665 |
+
"logps/chosen": -734.5894775390625,
|
| 666 |
+
"logps/rejected": -798.04296875,
|
| 667 |
+
"loss": 0.508,
|
| 668 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 669 |
+
"rewards/chosen": -4.4496073722839355,
|
| 670 |
+
"rewards/margins": 0.9138796925544739,
|
| 671 |
+
"rewards/rejected": -5.363487720489502,
|
| 672 |
+
"step": 400
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 0.837257980115123,
|
| 676 |
+
"eval_logits/chosen": -1.1871331930160522,
|
| 677 |
+
"eval_logits/rejected": -1.2109087705612183,
|
| 678 |
+
"eval_logps/chosen": -747.29638671875,
|
| 679 |
+
"eval_logps/rejected": -842.17724609375,
|
| 680 |
+
"eval_loss": 0.5157404541969299,
|
| 681 |
+
"eval_rewards/accuracies": 0.76953125,
|
| 682 |
+
"eval_rewards/chosen": -4.345571041107178,
|
| 683 |
+
"eval_rewards/margins": 1.0137183666229248,
|
| 684 |
+
"eval_rewards/rejected": -5.359289646148682,
|
| 685 |
+
"eval_runtime": 34.0708,
|
| 686 |
+
"eval_samples_per_second": 58.701,
|
| 687 |
+
"eval_steps_per_second": 0.939,
|
| 688 |
+
"step": 400
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.858189429618001,
|
| 692 |
+
"grad_norm": 0.22161860764026642,
|
| 693 |
+
"learning_rate": 1.7695632250191e-05,
|
| 694 |
+
"logits/chosen": -0.9348050951957703,
|
| 695 |
+
"logits/rejected": -1.0523372888565063,
|
| 696 |
+
"logps/chosen": -794.3521118164062,
|
| 697 |
+
"logps/rejected": -829.3026123046875,
|
| 698 |
+
"loss": 0.511,
|
| 699 |
+
"rewards/accuracies": 0.762499988079071,
|
| 700 |
+
"rewards/chosen": -4.382437705993652,
|
| 701 |
+
"rewards/margins": 0.990971565246582,
|
| 702 |
+
"rewards/rejected": -5.373409271240234,
|
| 703 |
+
"step": 410
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.8791208791208791,
|
| 707 |
+
"grad_norm": 0.2439228594303131,
|
| 708 |
+
"learning_rate": 1.2878971655412513e-05,
|
| 709 |
+
"logits/chosen": -1.1185680627822876,
|
| 710 |
+
"logits/rejected": -1.1273977756500244,
|
| 711 |
+
"logps/chosen": -739.8529052734375,
|
| 712 |
+
"logps/rejected": -834.0125122070312,
|
| 713 |
+
"loss": 0.4955,
|
| 714 |
+
"rewards/accuracies": 0.737500011920929,
|
| 715 |
+
"rewards/chosen": -4.466958045959473,
|
| 716 |
+
"rewards/margins": 0.9562638401985168,
|
| 717 |
+
"rewards/rejected": -5.423221588134766,
|
| 718 |
+
"step": 420
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.9000523286237572,
|
| 722 |
+
"grad_norm": 0.2282486855983734,
|
| 723 |
+
"learning_rate": 8.79732424714541e-06,
|
| 724 |
+
"logits/chosen": -1.2293227910995483,
|
| 725 |
+
"logits/rejected": -1.3410972356796265,
|
| 726 |
+
"logps/chosen": -786.1281127929688,
|
| 727 |
+
"logps/rejected": -832.2101440429688,
|
| 728 |
+
"loss": 0.4959,
|
| 729 |
+
"rewards/accuracies": 0.737500011920929,
|
| 730 |
+
"rewards/chosen": -4.519133567810059,
|
| 731 |
+
"rewards/margins": 0.8781270980834961,
|
| 732 |
+
"rewards/rejected": -5.397259712219238,
|
| 733 |
+
"step": 430
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 0.9209837781266352,
|
| 737 |
+
"grad_norm": 0.2305663526058197,
|
| 738 |
+
"learning_rate": 5.472568979361852e-06,
|
| 739 |
+
"logits/chosen": -1.0845152139663696,
|
| 740 |
+
"logits/rejected": -1.1872012615203857,
|
| 741 |
+
"logps/chosen": -724.5098266601562,
|
| 742 |
+
"logps/rejected": -755.2883911132812,
|
| 743 |
+
"loss": 0.519,
|
| 744 |
+
"rewards/accuracies": 0.737500011920929,
|
| 745 |
+
"rewards/chosen": -4.300398349761963,
|
| 746 |
+
"rewards/margins": 0.8024924397468567,
|
| 747 |
+
"rewards/rejected": -5.102890968322754,
|
| 748 |
+
"step": 440
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.9419152276295133,
|
| 752 |
+
"grad_norm": 0.22703427076339722,
|
| 753 |
+
"learning_rate": 2.922527618666465e-06,
|
| 754 |
+
"logits/chosen": -1.2591861486434937,
|
| 755 |
+
"logits/rejected": -1.3150806427001953,
|
| 756 |
+
"logps/chosen": -736.691650390625,
|
| 757 |
+
"logps/rejected": -812.67626953125,
|
| 758 |
+
"loss": 0.4926,
|
| 759 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 760 |
+
"rewards/chosen": -4.421994686126709,
|
| 761 |
+
"rewards/margins": 0.9103468060493469,
|
| 762 |
+
"rewards/rejected": -5.33234167098999,
|
| 763 |
+
"step": 450
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.9628466771323915,
|
| 767 |
+
"grad_norm": 0.25431472063064575,
|
| 768 |
+
"learning_rate": 1.1608692138469379e-06,
|
| 769 |
+
"logits/chosen": -1.1412612199783325,
|
| 770 |
+
"logits/rejected": -1.286179542541504,
|
| 771 |
+
"logps/chosen": -751.5174560546875,
|
| 772 |
+
"logps/rejected": -791.9276733398438,
|
| 773 |
+
"loss": 0.5078,
|
| 774 |
+
"rewards/accuracies": 0.762499988079071,
|
| 775 |
+
"rewards/chosen": -4.383217811584473,
|
| 776 |
+
"rewards/margins": 0.9881366491317749,
|
| 777 |
+
"rewards/rejected": -5.371354579925537,
|
| 778 |
+
"step": 460
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.9837781266352695,
|
| 782 |
+
"grad_norm": 0.2301347702741623,
|
| 783 |
+
"learning_rate": 1.9703682533901977e-07,
|
| 784 |
+
"logits/chosen": -1.1379841566085815,
|
| 785 |
+
"logits/rejected": -1.2217354774475098,
|
| 786 |
+
"logps/chosen": -767.3609619140625,
|
| 787 |
+
"logps/rejected": -820.5641479492188,
|
| 788 |
+
"loss": 0.4915,
|
| 789 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 790 |
+
"rewards/chosen": -4.3214111328125,
|
| 791 |
+
"rewards/margins": 0.8934388160705566,
|
| 792 |
+
"rewards/rejected": -5.214850425720215,
|
| 793 |
+
"step": 470
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.9984301412872841,
|
| 797 |
+
"step": 477,
|
| 798 |
+
"total_flos": 0.0,
|
| 799 |
+
"train_loss": 0.5423707397229016,
|
| 800 |
+
"train_runtime": 2137.6738,
|
| 801 |
+
"train_samples_per_second": 28.598,
|
| 802 |
+
"train_steps_per_second": 0.223
|
| 803 |
+
}
|
| 804 |
+
],
|
| 805 |
+
"logging_steps": 10,
|
| 806 |
+
"max_steps": 477,
|
| 807 |
+
"num_input_tokens_seen": 0,
|
| 808 |
+
"num_train_epochs": 1,
|
| 809 |
+
"save_steps": 100,
|
| 810 |
+
"stateful_callbacks": {
|
| 811 |
+
"TrainerControl": {
|
| 812 |
+
"args": {
|
| 813 |
+
"should_epoch_stop": false,
|
| 814 |
+
"should_evaluate": false,
|
| 815 |
+
"should_log": false,
|
| 816 |
+
"should_save": true,
|
| 817 |
+
"should_training_stop": true
|
| 818 |
+
},
|
| 819 |
+
"attributes": {}
|
| 820 |
+
}
|
| 821 |
+
},
|
| 822 |
+
"total_flos": 0.0,
|
| 823 |
+
"train_batch_size": 4,
|
| 824 |
+
"trial_name": null,
|
| 825 |
+
"trial_params": null
|
| 826 |
+
}
|