Model save
Browse files- README.md +77 -0
- adapter_model.safetensors +1 -1
- all_results.json +21 -0
- eval_results.json +16 -0
- runs/Jan19_12-34-55_08134be46a59/events.out.tfevents.1705667918.08134be46a59.4105.0 +2 -2
- runs/Jan19_12-34-55_08134be46a59/events.out.tfevents.1705685379.08134be46a59.4105.1 +3 -0
- train_results.json +8 -0
- trainer_state.json +1538 -0
README.md
ADDED
|
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
base_model: sablo/sablo-pebble-mistral
|
| 9 |
+
model-index:
|
| 10 |
+
- name: sablo-pebble-mistral-dpo-lora-HelpSteer_binarized-2
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# sablo-pebble-mistral-dpo-lora-HelpSteer_binarized-2
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [sablo/sablo-pebble-mistral](https://huggingface.co/sablo/sablo-pebble-mistral) on the None dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.5195
|
| 22 |
+
- Rewards/chosen: -1.3821
|
| 23 |
+
- Rewards/rejected: -2.4510
|
| 24 |
+
- Rewards/accuracies: 0.7358
|
| 25 |
+
- Rewards/margins: 1.0689
|
| 26 |
+
- Logps/rejected: -158.5470
|
| 27 |
+
- Logps/chosen: -147.7195
|
| 28 |
+
- Logits/rejected: -2.0952
|
| 29 |
+
- Logits/chosen: -2.1922
|
| 30 |
+
|
| 31 |
+
## Model description
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Intended uses & limitations
|
| 36 |
+
|
| 37 |
+
More information needed
|
| 38 |
+
|
| 39 |
+
## Training and evaluation data
|
| 40 |
+
|
| 41 |
+
More information needed
|
| 42 |
+
|
| 43 |
+
## Training procedure
|
| 44 |
+
|
| 45 |
+
### Training hyperparameters
|
| 46 |
+
|
| 47 |
+
The following hyperparameters were used during training:
|
| 48 |
+
- learning_rate: 5e-06
|
| 49 |
+
- train_batch_size: 4
|
| 50 |
+
- eval_batch_size: 8
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- gradient_accumulation_steps: 2
|
| 54 |
+
- total_train_batch_size: 8
|
| 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 |
+
| 0.65 | 0.2 | 200 | 0.6563 | 0.1070 | 0.0177 | 0.6509 | 0.0893 | -76.2561 | -98.0835 | -2.0464 | -2.1421 |
|
| 65 |
+
| 0.456 | 0.39 | 400 | 0.5446 | -1.2305 | -1.8748 | 0.7217 | 0.6444 | -139.3410 | -142.6661 | -2.1203 | -2.2102 |
|
| 66 |
+
| 0.4388 | 0.59 | 600 | 0.5325 | -1.8012 | -2.8927 | 0.7123 | 1.0915 | -173.2708 | -161.6904 | -2.1017 | -2.1954 |
|
| 67 |
+
| 0.6137 | 0.79 | 800 | 0.5198 | -1.4487 | -2.5199 | 0.7382 | 1.0712 | -160.8413 | -149.9388 | -2.0962 | -2.1935 |
|
| 68 |
+
| 0.5866 | 0.98 | 1000 | 0.5195 | -1.3821 | -2.4510 | 0.7358 | 1.0689 | -158.5470 | -147.7195 | -2.0952 | -2.1922 |
|
| 69 |
+
|
| 70 |
+
|
| 71 |
+
### Framework versions
|
| 72 |
+
|
| 73 |
+
- PEFT 0.7.1
|
| 74 |
+
- Transformers 4.36.2
|
| 75 |
+
- Pytorch 2.0.1+cu118
|
| 76 |
+
- Datasets 2.14.6
|
| 77 |
+
- Tokenizers 0.15.0
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 83945744
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:48f734b7073eba57a580259f094f177adf210cc855e0e23332611cf08a9c73f9
|
| 3 |
size 83945744
|
all_results.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_logits/chosen": -2.192172050476074,
|
| 4 |
+
"eval_logits/rejected": -2.095205545425415,
|
| 5 |
+
"eval_logps/chosen": -147.71951293945312,
|
| 6 |
+
"eval_logps/rejected": -158.54696655273438,
|
| 7 |
+
"eval_loss": 0.5194684863090515,
|
| 8 |
+
"eval_rewards/accuracies": 0.7358490824699402,
|
| 9 |
+
"eval_rewards/chosen": -1.382088541984558,
|
| 10 |
+
"eval_rewards/margins": 1.068930983543396,
|
| 11 |
+
"eval_rewards/rejected": -2.451019525527954,
|
| 12 |
+
"eval_runtime": 422.7886,
|
| 13 |
+
"eval_samples": 418,
|
| 14 |
+
"eval_samples_per_second": 0.989,
|
| 15 |
+
"eval_steps_per_second": 0.125,
|
| 16 |
+
"train_loss": 0.5536013367607837,
|
| 17 |
+
"train_runtime": 17040.3461,
|
| 18 |
+
"train_samples": 8130,
|
| 19 |
+
"train_samples_per_second": 0.477,
|
| 20 |
+
"train_steps_per_second": 0.06
|
| 21 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_logits/chosen": -2.192172050476074,
|
| 4 |
+
"eval_logits/rejected": -2.095205545425415,
|
| 5 |
+
"eval_logps/chosen": -147.71951293945312,
|
| 6 |
+
"eval_logps/rejected": -158.54696655273438,
|
| 7 |
+
"eval_loss": 0.5194684863090515,
|
| 8 |
+
"eval_rewards/accuracies": 0.7358490824699402,
|
| 9 |
+
"eval_rewards/chosen": -1.382088541984558,
|
| 10 |
+
"eval_rewards/margins": 1.068930983543396,
|
| 11 |
+
"eval_rewards/rejected": -2.451019525527954,
|
| 12 |
+
"eval_runtime": 422.7886,
|
| 13 |
+
"eval_samples": 418,
|
| 14 |
+
"eval_samples_per_second": 0.989,
|
| 15 |
+
"eval_steps_per_second": 0.125
|
| 16 |
+
}
|
runs/Jan19_12-34-55_08134be46a59/events.out.tfevents.1705667918.08134be46a59.4105.0
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:b7981d59edf7b04164e13033db557c7a6707655e9adf64dc75f8da6c3d41de20
|
| 3 |
+
size 73263
|
runs/Jan19_12-34-55_08134be46a59/events.out.tfevents.1705685379.08134be46a59.4105.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bffeef1461d345b389ec89af41f16d731602330205245d5772927c7a2ed597fd
|
| 3 |
+
size 828
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.5536013367607837,
|
| 4 |
+
"train_runtime": 17040.3461,
|
| 5 |
+
"train_samples": 8130,
|
| 6 |
+
"train_samples_per_second": 0.477,
|
| 7 |
+
"train_steps_per_second": 0.06
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1538 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.999508116084604,
|
| 5 |
+
"eval_steps": 200,
|
| 6 |
+
"global_step": 1016,
|
| 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": 4.901960784313726e-08,
|
| 14 |
+
"logits/chosen": -2.0748214721679688,
|
| 15 |
+
"logits/rejected": -2.1528916358947754,
|
| 16 |
+
"logps/chosen": -78.83467102050781,
|
| 17 |
+
"logps/rejected": -90.04127502441406,
|
| 18 |
+
"loss": 0.6931,
|
| 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": 0.01,
|
| 27 |
+
"learning_rate": 4.901960784313725e-07,
|
| 28 |
+
"logits/chosen": -2.167158603668213,
|
| 29 |
+
"logits/rejected": -2.065119743347168,
|
| 30 |
+
"logps/chosen": -101.462646484375,
|
| 31 |
+
"logps/rejected": -67.91659545898438,
|
| 32 |
+
"loss": 0.6937,
|
| 33 |
+
"rewards/accuracies": 0.3611111044883728,
|
| 34 |
+
"rewards/chosen": -0.0003858842537738383,
|
| 35 |
+
"rewards/margins": -0.001147754956036806,
|
| 36 |
+
"rewards/rejected": 0.0007618705276399851,
|
| 37 |
+
"step": 10
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.02,
|
| 41 |
+
"learning_rate": 9.80392156862745e-07,
|
| 42 |
+
"logits/chosen": -2.139626979827881,
|
| 43 |
+
"logits/rejected": -2.0195488929748535,
|
| 44 |
+
"logps/chosen": -109.9924545288086,
|
| 45 |
+
"logps/rejected": -84.65155029296875,
|
| 46 |
+
"loss": 0.695,
|
| 47 |
+
"rewards/accuracies": 0.375,
|
| 48 |
+
"rewards/chosen": -0.002081229817122221,
|
| 49 |
+
"rewards/margins": -0.0036208624951541424,
|
| 50 |
+
"rewards/rejected": 0.0015396325616165996,
|
| 51 |
+
"step": 20
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.03,
|
| 55 |
+
"learning_rate": 1.4705882352941177e-06,
|
| 56 |
+
"logits/chosen": -2.2819817066192627,
|
| 57 |
+
"logits/rejected": -2.181386947631836,
|
| 58 |
+
"logps/chosen": -100.8692398071289,
|
| 59 |
+
"logps/rejected": -84.57305908203125,
|
| 60 |
+
"loss": 0.6931,
|
| 61 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 62 |
+
"rewards/chosen": 0.0019000970060005784,
|
| 63 |
+
"rewards/margins": 0.00016770660295151174,
|
| 64 |
+
"rewards/rejected": 0.0017323906067758799,
|
| 65 |
+
"step": 30
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.04,
|
| 69 |
+
"learning_rate": 1.96078431372549e-06,
|
| 70 |
+
"logits/chosen": -2.274721622467041,
|
| 71 |
+
"logits/rejected": -2.156294584274292,
|
| 72 |
+
"logps/chosen": -105.41255950927734,
|
| 73 |
+
"logps/rejected": -83.99583435058594,
|
| 74 |
+
"loss": 0.6929,
|
| 75 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 76 |
+
"rewards/chosen": 0.001408256939612329,
|
| 77 |
+
"rewards/margins": 0.0004250857455190271,
|
| 78 |
+
"rewards/rejected": 0.0009831712814047933,
|
| 79 |
+
"step": 40
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.05,
|
| 83 |
+
"learning_rate": 2.450980392156863e-06,
|
| 84 |
+
"logits/chosen": -2.29875111579895,
|
| 85 |
+
"logits/rejected": -2.20742130279541,
|
| 86 |
+
"logps/chosen": -103.51786041259766,
|
| 87 |
+
"logps/rejected": -77.90748596191406,
|
| 88 |
+
"loss": 0.692,
|
| 89 |
+
"rewards/accuracies": 0.550000011920929,
|
| 90 |
+
"rewards/chosen": 0.005386353936046362,
|
| 91 |
+
"rewards/margins": 0.002320895902812481,
|
| 92 |
+
"rewards/rejected": 0.0030654582660645247,
|
| 93 |
+
"step": 50
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.06,
|
| 97 |
+
"learning_rate": 2.9411764705882355e-06,
|
| 98 |
+
"logits/chosen": -2.30989408493042,
|
| 99 |
+
"logits/rejected": -2.233097553253174,
|
| 100 |
+
"logps/chosen": -107.9452896118164,
|
| 101 |
+
"logps/rejected": -90.55638122558594,
|
| 102 |
+
"loss": 0.6912,
|
| 103 |
+
"rewards/accuracies": 0.637499988079071,
|
| 104 |
+
"rewards/chosen": 0.008994068950414658,
|
| 105 |
+
"rewards/margins": 0.003984561190009117,
|
| 106 |
+
"rewards/rejected": 0.005009507294744253,
|
| 107 |
+
"step": 60
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.07,
|
| 111 |
+
"learning_rate": 3.431372549019608e-06,
|
| 112 |
+
"logits/chosen": -2.2964837551116943,
|
| 113 |
+
"logits/rejected": -2.192676544189453,
|
| 114 |
+
"logps/chosen": -98.52997589111328,
|
| 115 |
+
"logps/rejected": -73.2212142944336,
|
| 116 |
+
"loss": 0.6907,
|
| 117 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 118 |
+
"rewards/chosen": 0.011149758473038673,
|
| 119 |
+
"rewards/margins": 0.004859681706875563,
|
| 120 |
+
"rewards/rejected": 0.006290078163146973,
|
| 121 |
+
"step": 70
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.08,
|
| 125 |
+
"learning_rate": 3.92156862745098e-06,
|
| 126 |
+
"logits/chosen": -2.147590398788452,
|
| 127 |
+
"logits/rejected": -2.0242252349853516,
|
| 128 |
+
"logps/chosen": -110.5780258178711,
|
| 129 |
+
"logps/rejected": -85.19944763183594,
|
| 130 |
+
"loss": 0.6903,
|
| 131 |
+
"rewards/accuracies": 0.612500011920929,
|
| 132 |
+
"rewards/chosen": 0.018292848020792007,
|
| 133 |
+
"rewards/margins": 0.005854339338839054,
|
| 134 |
+
"rewards/rejected": 0.012438507750630379,
|
| 135 |
+
"step": 80
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.09,
|
| 139 |
+
"learning_rate": 4.411764705882353e-06,
|
| 140 |
+
"logits/chosen": -2.103201389312744,
|
| 141 |
+
"logits/rejected": -1.9731849431991577,
|
| 142 |
+
"logps/chosen": -86.84721374511719,
|
| 143 |
+
"logps/rejected": -63.73047637939453,
|
| 144 |
+
"loss": 0.6889,
|
| 145 |
+
"rewards/accuracies": 0.625,
|
| 146 |
+
"rewards/chosen": 0.02545500174164772,
|
| 147 |
+
"rewards/margins": 0.008647125214338303,
|
| 148 |
+
"rewards/rejected": 0.016807876527309418,
|
| 149 |
+
"step": 90
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.1,
|
| 153 |
+
"learning_rate": 4.901960784313726e-06,
|
| 154 |
+
"logits/chosen": -2.071173906326294,
|
| 155 |
+
"logits/rejected": -1.8952083587646484,
|
| 156 |
+
"logps/chosen": -120.53657531738281,
|
| 157 |
+
"logps/rejected": -84.6740951538086,
|
| 158 |
+
"loss": 0.6804,
|
| 159 |
+
"rewards/accuracies": 0.75,
|
| 160 |
+
"rewards/chosen": 0.051459141075611115,
|
| 161 |
+
"rewards/margins": 0.02619524672627449,
|
| 162 |
+
"rewards/rejected": 0.025263899937272072,
|
| 163 |
+
"step": 100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.11,
|
| 167 |
+
"learning_rate": 4.9990549169459415e-06,
|
| 168 |
+
"logits/chosen": -2.2821316719055176,
|
| 169 |
+
"logits/rejected": -2.1455578804016113,
|
| 170 |
+
"logps/chosen": -103.51020812988281,
|
| 171 |
+
"logps/rejected": -78.41320037841797,
|
| 172 |
+
"loss": 0.6767,
|
| 173 |
+
"rewards/accuracies": 0.824999988079071,
|
| 174 |
+
"rewards/chosen": 0.06272837519645691,
|
| 175 |
+
"rewards/margins": 0.03374689072370529,
|
| 176 |
+
"rewards/rejected": 0.028981482610106468,
|
| 177 |
+
"step": 110
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.12,
|
| 181 |
+
"learning_rate": 4.995216741642263e-06,
|
| 182 |
+
"logits/chosen": -2.284080743789673,
|
| 183 |
+
"logits/rejected": -2.210580825805664,
|
| 184 |
+
"logps/chosen": -96.24214172363281,
|
| 185 |
+
"logps/rejected": -77.19042205810547,
|
| 186 |
+
"loss": 0.6812,
|
| 187 |
+
"rewards/accuracies": 0.612500011920929,
|
| 188 |
+
"rewards/chosen": 0.08176051080226898,
|
| 189 |
+
"rewards/margins": 0.02550395205616951,
|
| 190 |
+
"rewards/rejected": 0.056256555020809174,
|
| 191 |
+
"step": 120
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.13,
|
| 195 |
+
"learning_rate": 4.988430936991089e-06,
|
| 196 |
+
"logits/chosen": -2.2925381660461426,
|
| 197 |
+
"logits/rejected": -2.149710178375244,
|
| 198 |
+
"logps/chosen": -106.36643981933594,
|
| 199 |
+
"logps/rejected": -80.01815795898438,
|
| 200 |
+
"loss": 0.6647,
|
| 201 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 202 |
+
"rewards/chosen": 0.11486251652240753,
|
| 203 |
+
"rewards/margins": 0.06016199663281441,
|
| 204 |
+
"rewards/rejected": 0.054700516164302826,
|
| 205 |
+
"step": 130
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.14,
|
| 209 |
+
"learning_rate": 4.978705519144525e-06,
|
| 210 |
+
"logits/chosen": -2.1970953941345215,
|
| 211 |
+
"logits/rejected": -2.0385475158691406,
|
| 212 |
+
"logps/chosen": -118.03541564941406,
|
| 213 |
+
"logps/rejected": -81.73417663574219,
|
| 214 |
+
"loss": 0.6639,
|
| 215 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 216 |
+
"rewards/chosen": 0.10544681549072266,
|
| 217 |
+
"rewards/margins": 0.06229867413640022,
|
| 218 |
+
"rewards/rejected": 0.04314813390374184,
|
| 219 |
+
"step": 140
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.15,
|
| 223 |
+
"learning_rate": 4.966051976854862e-06,
|
| 224 |
+
"logits/chosen": -2.339622735977173,
|
| 225 |
+
"logits/rejected": -2.2239785194396973,
|
| 226 |
+
"logps/chosen": -90.04592895507812,
|
| 227 |
+
"logps/rejected": -68.85932922363281,
|
| 228 |
+
"loss": 0.6506,
|
| 229 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 230 |
+
"rewards/chosen": 0.1409004032611847,
|
| 231 |
+
"rewards/margins": 0.09264969825744629,
|
| 232 |
+
"rewards/rejected": 0.048250701278448105,
|
| 233 |
+
"step": 150
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.16,
|
| 237 |
+
"learning_rate": 4.950485257902782e-06,
|
| 238 |
+
"logits/chosen": -2.2208049297332764,
|
| 239 |
+
"logits/rejected": -2.1466565132141113,
|
| 240 |
+
"logps/chosen": -101.00809478759766,
|
| 241 |
+
"logps/rejected": -76.30353546142578,
|
| 242 |
+
"loss": 0.6618,
|
| 243 |
+
"rewards/accuracies": 0.675000011920929,
|
| 244 |
+
"rewards/chosen": 0.13603894412517548,
|
| 245 |
+
"rewards/margins": 0.0690341368317604,
|
| 246 |
+
"rewards/rejected": 0.06700481474399567,
|
| 247 |
+
"step": 160
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.17,
|
| 251 |
+
"learning_rate": 4.932023751439358e-06,
|
| 252 |
+
"logits/chosen": -2.286597728729248,
|
| 253 |
+
"logits/rejected": -2.13307523727417,
|
| 254 |
+
"logps/chosen": -110.24186706542969,
|
| 255 |
+
"logps/rejected": -81.82392120361328,
|
| 256 |
+
"loss": 0.6666,
|
| 257 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 258 |
+
"rewards/chosen": 0.13935917615890503,
|
| 259 |
+
"rewards/margins": 0.06269019842147827,
|
| 260 |
+
"rewards/rejected": 0.07666899263858795,
|
| 261 |
+
"step": 170
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.18,
|
| 265 |
+
"learning_rate": 4.9106892662627395e-06,
|
| 266 |
+
"logits/chosen": -2.357682466506958,
|
| 267 |
+
"logits/rejected": -2.237700939178467,
|
| 268 |
+
"logps/chosen": -104.79756927490234,
|
| 269 |
+
"logps/rejected": -81.48443603515625,
|
| 270 |
+
"loss": 0.6615,
|
| 271 |
+
"rewards/accuracies": 0.6875,
|
| 272 |
+
"rewards/chosen": 0.16287028789520264,
|
| 273 |
+
"rewards/margins": 0.07204125821590424,
|
| 274 |
+
"rewards/rejected": 0.0908290296792984,
|
| 275 |
+
"step": 180
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.19,
|
| 279 |
+
"learning_rate": 4.886507005055149e-06,
|
| 280 |
+
"logits/chosen": -2.3158655166625977,
|
| 281 |
+
"logits/rejected": -2.1448147296905518,
|
| 282 |
+
"logps/chosen": -113.81204986572266,
|
| 283 |
+
"logps/rejected": -80.2816390991211,
|
| 284 |
+
"loss": 0.646,
|
| 285 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 286 |
+
"rewards/chosen": 0.15469492971897125,
|
| 287 |
+
"rewards/margins": 0.10541819036006927,
|
| 288 |
+
"rewards/rejected": 0.049276743084192276,
|
| 289 |
+
"step": 190
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.2,
|
| 293 |
+
"learning_rate": 4.859505534610658e-06,
|
| 294 |
+
"logits/chosen": -2.2734456062316895,
|
| 295 |
+
"logits/rejected": -2.1840872764587402,
|
| 296 |
+
"logps/chosen": -94.27523040771484,
|
| 297 |
+
"logps/rejected": -76.13407897949219,
|
| 298 |
+
"loss": 0.65,
|
| 299 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 300 |
+
"rewards/chosen": 0.12800325453281403,
|
| 301 |
+
"rewards/margins": 0.10445593297481537,
|
| 302 |
+
"rewards/rejected": 0.023547304794192314,
|
| 303 |
+
"step": 200
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.2,
|
| 307 |
+
"eval_logits/chosen": -2.1420769691467285,
|
| 308 |
+
"eval_logits/rejected": -2.046384811401367,
|
| 309 |
+
"eval_logps/chosen": -98.08348083496094,
|
| 310 |
+
"eval_logps/rejected": -76.25611877441406,
|
| 311 |
+
"eval_loss": 0.656255841255188,
|
| 312 |
+
"eval_rewards/accuracies": 0.650943398475647,
|
| 313 |
+
"eval_rewards/chosen": 0.106992706656456,
|
| 314 |
+
"eval_rewards/margins": 0.08928683400154114,
|
| 315 |
+
"eval_rewards/rejected": 0.017705870792269707,
|
| 316 |
+
"eval_runtime": 425.8967,
|
| 317 |
+
"eval_samples_per_second": 0.981,
|
| 318 |
+
"eval_steps_per_second": 0.124,
|
| 319 |
+
"step": 200
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.21,
|
| 323 |
+
"learning_rate": 4.829716752088893e-06,
|
| 324 |
+
"logits/chosen": -2.2180724143981934,
|
| 325 |
+
"logits/rejected": -2.1603446006774902,
|
| 326 |
+
"logps/chosen": -85.03736877441406,
|
| 327 |
+
"logps/rejected": -78.87088012695312,
|
| 328 |
+
"loss": 0.6573,
|
| 329 |
+
"rewards/accuracies": 0.699999988079071,
|
| 330 |
+
"rewards/chosen": 0.10240446031093597,
|
| 331 |
+
"rewards/margins": 0.08565853536128998,
|
| 332 |
+
"rewards/rejected": 0.016745951026678085,
|
| 333 |
+
"step": 210
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.22,
|
| 337 |
+
"learning_rate": 4.797175847334535e-06,
|
| 338 |
+
"logits/chosen": -2.2404112815856934,
|
| 339 |
+
"logits/rejected": -2.12317156791687,
|
| 340 |
+
"logps/chosen": -107.25093841552734,
|
| 341 |
+
"logps/rejected": -84.33776092529297,
|
| 342 |
+
"loss": 0.6679,
|
| 343 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 344 |
+
"rewards/chosen": 0.06942565739154816,
|
| 345 |
+
"rewards/margins": 0.06278938055038452,
|
| 346 |
+
"rewards/rejected": 0.006636275444179773,
|
| 347 |
+
"step": 220
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.23,
|
| 351 |
+
"learning_rate": 4.761921261307143e-06,
|
| 352 |
+
"logits/chosen": -2.271301746368408,
|
| 353 |
+
"logits/rejected": -2.1368188858032227,
|
| 354 |
+
"logps/chosen": -102.20993041992188,
|
| 355 |
+
"logps/rejected": -80.41031646728516,
|
| 356 |
+
"loss": 0.656,
|
| 357 |
+
"rewards/accuracies": 0.699999988079071,
|
| 358 |
+
"rewards/chosen": 0.06752457469701767,
|
| 359 |
+
"rewards/margins": 0.09335462749004364,
|
| 360 |
+
"rewards/rejected": -0.02583005093038082,
|
| 361 |
+
"step": 230
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.24,
|
| 365 |
+
"learning_rate": 4.723994640670377e-06,
|
| 366 |
+
"logits/chosen": -2.2616968154907227,
|
| 367 |
+
"logits/rejected": -2.090630292892456,
|
| 368 |
+
"logps/chosen": -112.83064270019531,
|
| 369 |
+
"logps/rejected": -78.7665023803711,
|
| 370 |
+
"loss": 0.613,
|
| 371 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 372 |
+
"rewards/chosen": 0.1038433164358139,
|
| 373 |
+
"rewards/margins": 0.18221423029899597,
|
| 374 |
+
"rewards/rejected": -0.07837091386318207,
|
| 375 |
+
"step": 240
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.25,
|
| 379 |
+
"learning_rate": 4.68344078859431e-06,
|
| 380 |
+
"logits/chosen": -2.214836597442627,
|
| 381 |
+
"logits/rejected": -2.1651015281677246,
|
| 382 |
+
"logps/chosen": -94.0736312866211,
|
| 383 |
+
"logps/rejected": -87.97615814208984,
|
| 384 |
+
"loss": 0.6446,
|
| 385 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 386 |
+
"rewards/chosen": 0.05433257669210434,
|
| 387 |
+
"rewards/margins": 0.1173890084028244,
|
| 388 |
+
"rewards/rejected": -0.06305641680955887,
|
| 389 |
+
"step": 250
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.26,
|
| 393 |
+
"learning_rate": 4.6403076118289006e-06,
|
| 394 |
+
"logits/chosen": -2.191092014312744,
|
| 395 |
+
"logits/rejected": -2.0465681552886963,
|
| 396 |
+
"logps/chosen": -106.81085205078125,
|
| 397 |
+
"logps/rejected": -80.91065216064453,
|
| 398 |
+
"loss": 0.6325,
|
| 399 |
+
"rewards/accuracies": 0.675000011920929,
|
| 400 |
+
"rewards/chosen": 0.04650095850229263,
|
| 401 |
+
"rewards/margins": 0.14686846733093262,
|
| 402 |
+
"rewards/rejected": -0.10036750882863998,
|
| 403 |
+
"step": 260
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.27,
|
| 407 |
+
"learning_rate": 4.5946460641111776e-06,
|
| 408 |
+
"logits/chosen": -2.3747944831848145,
|
| 409 |
+
"logits/rejected": -2.1630542278289795,
|
| 410 |
+
"logps/chosen": -103.29341125488281,
|
| 411 |
+
"logps/rejected": -80.09381103515625,
|
| 412 |
+
"loss": 0.6041,
|
| 413 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 414 |
+
"rewards/chosen": 0.06812257319688797,
|
| 415 |
+
"rewards/margins": 0.2125866860151291,
|
| 416 |
+
"rewards/rejected": -0.14446412026882172,
|
| 417 |
+
"step": 270
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.28,
|
| 421 |
+
"learning_rate": 4.546510085972983e-06,
|
| 422 |
+
"logits/chosen": -2.385852813720703,
|
| 423 |
+
"logits/rejected": -2.2712764739990234,
|
| 424 |
+
"logps/chosen": -115.0221176147461,
|
| 425 |
+
"logps/rejected": -90.1436538696289,
|
| 426 |
+
"loss": 0.6212,
|
| 427 |
+
"rewards/accuracies": 0.675000011920929,
|
| 428 |
+
"rewards/chosen": 0.017763856798410416,
|
| 429 |
+
"rewards/margins": 0.1770613044500351,
|
| 430 |
+
"rewards/rejected": -0.15929746627807617,
|
| 431 |
+
"step": 280
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.29,
|
| 435 |
+
"learning_rate": 4.495956541020376e-06,
|
| 436 |
+
"logits/chosen": -2.3901445865631104,
|
| 437 |
+
"logits/rejected": -2.247413396835327,
|
| 438 |
+
"logps/chosen": -119.44215393066406,
|
| 439 |
+
"logps/rejected": -101.89396667480469,
|
| 440 |
+
"loss": 0.613,
|
| 441 |
+
"rewards/accuracies": 0.625,
|
| 442 |
+
"rewards/chosen": -0.03466514125466347,
|
| 443 |
+
"rewards/margins": 0.2126956284046173,
|
| 444 |
+
"rewards/rejected": -0.24736078083515167,
|
| 445 |
+
"step": 290
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.3,
|
| 449 |
+
"learning_rate": 4.443045148759978e-06,
|
| 450 |
+
"logits/chosen": -2.3851912021636963,
|
| 451 |
+
"logits/rejected": -2.246699094772339,
|
| 452 |
+
"logps/chosen": -126.57054138183594,
|
| 453 |
+
"logps/rejected": -93.43769836425781,
|
| 454 |
+
"loss": 0.5898,
|
| 455 |
+
"rewards/accuracies": 0.6875,
|
| 456 |
+
"rewards/chosen": -0.06855957210063934,
|
| 457 |
+
"rewards/margins": 0.2794524133205414,
|
| 458 |
+
"rewards/rejected": -0.3480120301246643,
|
| 459 |
+
"step": 300
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.3,
|
| 463 |
+
"learning_rate": 4.3878384140516025e-06,
|
| 464 |
+
"logits/chosen": -2.4334146976470947,
|
| 465 |
+
"logits/rejected": -2.3179290294647217,
|
| 466 |
+
"logps/chosen": -118.78206634521484,
|
| 467 |
+
"logps/rejected": -107.4043197631836,
|
| 468 |
+
"loss": 0.5792,
|
| 469 |
+
"rewards/accuracies": 0.699999988079071,
|
| 470 |
+
"rewards/chosen": -0.15513427555561066,
|
| 471 |
+
"rewards/margins": 0.30940642952919006,
|
| 472 |
+
"rewards/rejected": -0.46454063057899475,
|
| 473 |
+
"step": 310
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.31,
|
| 477 |
+
"learning_rate": 4.330401553270522e-06,
|
| 478 |
+
"logits/chosen": -2.34911847114563,
|
| 479 |
+
"logits/rejected": -2.2519352436065674,
|
| 480 |
+
"logps/chosen": -116.80616760253906,
|
| 481 |
+
"logps/rejected": -108.0991439819336,
|
| 482 |
+
"loss": 0.5869,
|
| 483 |
+
"rewards/accuracies": 0.737500011920929,
|
| 484 |
+
"rewards/chosen": -0.32395830750465393,
|
| 485 |
+
"rewards/margins": 0.30500033497810364,
|
| 486 |
+
"rewards/rejected": -0.6289585828781128,
|
| 487 |
+
"step": 320
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 0.32,
|
| 491 |
+
"learning_rate": 4.2708024172665795e-06,
|
| 492 |
+
"logits/chosen": -2.448741912841797,
|
| 493 |
+
"logits/rejected": -2.2790164947509766,
|
| 494 |
+
"logps/chosen": -119.88993072509766,
|
| 495 |
+
"logps/rejected": -94.18651580810547,
|
| 496 |
+
"loss": 0.5564,
|
| 497 |
+
"rewards/accuracies": 0.8125,
|
| 498 |
+
"rewards/chosen": -0.37532466650009155,
|
| 499 |
+
"rewards/margins": 0.37958258390426636,
|
| 500 |
+
"rewards/rejected": -0.7549072504043579,
|
| 501 |
+
"step": 330
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.33,
|
| 505 |
+
"learning_rate": 4.209111411211174e-06,
|
| 506 |
+
"logits/chosen": -2.4606077671051025,
|
| 507 |
+
"logits/rejected": -2.3158764839172363,
|
| 508 |
+
"logps/chosen": -108.91439056396484,
|
| 509 |
+
"logps/rejected": -94.19551086425781,
|
| 510 |
+
"loss": 0.5416,
|
| 511 |
+
"rewards/accuracies": 0.75,
|
| 512 |
+
"rewards/chosen": -0.3131277859210968,
|
| 513 |
+
"rewards/margins": 0.4208511710166931,
|
| 514 |
+
"rewards/rejected": -0.7339790463447571,
|
| 515 |
+
"step": 340
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.34,
|
| 519 |
+
"learning_rate": 4.145401411426788e-06,
|
| 520 |
+
"logits/chosen": -2.4888482093811035,
|
| 521 |
+
"logits/rejected": -2.3739964962005615,
|
| 522 |
+
"logps/chosen": -106.79335021972656,
|
| 523 |
+
"logps/rejected": -94.10002136230469,
|
| 524 |
+
"loss": 0.5863,
|
| 525 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 526 |
+
"rewards/chosen": -0.24795857071876526,
|
| 527 |
+
"rewards/margins": 0.3726752996444702,
|
| 528 |
+
"rewards/rejected": -0.6206337809562683,
|
| 529 |
+
"step": 350
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 0.35,
|
| 533 |
+
"learning_rate": 4.079747679297314e-06,
|
| 534 |
+
"logits/chosen": -2.320113182067871,
|
| 535 |
+
"logits/rejected": -2.214874744415283,
|
| 536 |
+
"logps/chosen": -118.53324890136719,
|
| 537 |
+
"logps/rejected": -100.85164642333984,
|
| 538 |
+
"loss": 0.574,
|
| 539 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 540 |
+
"rewards/chosen": -0.29153329133987427,
|
| 541 |
+
"rewards/margins": 0.3650625944137573,
|
| 542 |
+
"rewards/rejected": -0.6565959453582764,
|
| 543 |
+
"step": 360
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.36,
|
| 547 |
+
"learning_rate": 4.012227772360889e-06,
|
| 548 |
+
"logits/chosen": -2.3389956951141357,
|
| 549 |
+
"logits/rejected": -2.1818504333496094,
|
| 550 |
+
"logps/chosen": -123.92159271240234,
|
| 551 |
+
"logps/rejected": -110.42405700683594,
|
| 552 |
+
"loss": 0.5134,
|
| 553 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 554 |
+
"rewards/chosen": -0.2689405083656311,
|
| 555 |
+
"rewards/margins": 0.5353353023529053,
|
| 556 |
+
"rewards/rejected": -0.8042758703231812,
|
| 557 |
+
"step": 370
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.37,
|
| 561 |
+
"learning_rate": 3.942921452690245e-06,
|
| 562 |
+
"logits/chosen": -2.369523525238037,
|
| 563 |
+
"logits/rejected": -2.285311222076416,
|
| 564 |
+
"logps/chosen": -132.2254180908203,
|
| 565 |
+
"logps/rejected": -125.16448974609375,
|
| 566 |
+
"loss": 0.6273,
|
| 567 |
+
"rewards/accuracies": 0.699999988079071,
|
| 568 |
+
"rewards/chosen": -0.6799182891845703,
|
| 569 |
+
"rewards/margins": 0.3109356760978699,
|
| 570 |
+
"rewards/rejected": -0.9908539652824402,
|
| 571 |
+
"step": 380
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 0.38,
|
| 575 |
+
"learning_rate": 3.871910592668817e-06,
|
| 576 |
+
"logits/chosen": -2.4459290504455566,
|
| 577 |
+
"logits/rejected": -2.3258919715881348,
|
| 578 |
+
"logps/chosen": -127.33780670166016,
|
| 579 |
+
"logps/rejected": -129.97348022460938,
|
| 580 |
+
"loss": 0.4728,
|
| 581 |
+
"rewards/accuracies": 0.800000011920929,
|
| 582 |
+
"rewards/chosen": -0.7220331430435181,
|
| 583 |
+
"rewards/margins": 0.7715710997581482,
|
| 584 |
+
"rewards/rejected": -1.4936041831970215,
|
| 585 |
+
"step": 390
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.39,
|
| 589 |
+
"learning_rate": 3.799279078273921e-06,
|
| 590 |
+
"logits/chosen": -2.3586697578430176,
|
| 591 |
+
"logits/rejected": -2.190298080444336,
|
| 592 |
+
"logps/chosen": -137.87445068359375,
|
| 593 |
+
"logps/rejected": -125.60152435302734,
|
| 594 |
+
"loss": 0.456,
|
| 595 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 596 |
+
"rewards/chosen": -0.9262881278991699,
|
| 597 |
+
"rewards/margins": 0.8406116366386414,
|
| 598 |
+
"rewards/rejected": -1.766899824142456,
|
| 599 |
+
"step": 400
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 0.39,
|
| 603 |
+
"eval_logits/chosen": -2.21016526222229,
|
| 604 |
+
"eval_logits/rejected": -2.1202762126922607,
|
| 605 |
+
"eval_logps/chosen": -142.6661376953125,
|
| 606 |
+
"eval_logps/rejected": -139.3409881591797,
|
| 607 |
+
"eval_loss": 0.5446351766586304,
|
| 608 |
+
"eval_rewards/accuracies": 0.7216981053352356,
|
| 609 |
+
"eval_rewards/chosen": -1.23048734664917,
|
| 610 |
+
"eval_rewards/margins": 0.6443533301353455,
|
| 611 |
+
"eval_rewards/rejected": -1.8748407363891602,
|
| 612 |
+
"eval_runtime": 427.1638,
|
| 613 |
+
"eval_samples_per_second": 0.979,
|
| 614 |
+
"eval_steps_per_second": 0.124,
|
| 615 |
+
"step": 400
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.4,
|
| 619 |
+
"learning_rate": 3.725112709981249e-06,
|
| 620 |
+
"logits/chosen": -2.2144548892974854,
|
| 621 |
+
"logits/rejected": -2.0825304985046387,
|
| 622 |
+
"logps/chosen": -153.91371154785156,
|
| 623 |
+
"logps/rejected": -152.90274047851562,
|
| 624 |
+
"loss": 0.524,
|
| 625 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 626 |
+
"rewards/chosen": -1.2480976581573486,
|
| 627 |
+
"rewards/margins": 0.7566108703613281,
|
| 628 |
+
"rewards/rejected": -2.0047080516815186,
|
| 629 |
+
"step": 410
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 0.41,
|
| 633 |
+
"learning_rate": 3.649499101407737e-06,
|
| 634 |
+
"logits/chosen": -2.3662781715393066,
|
| 635 |
+
"logits/rejected": -2.2161378860473633,
|
| 636 |
+
"logps/chosen": -157.3763427734375,
|
| 637 |
+
"logps/rejected": -154.7397918701172,
|
| 638 |
+
"loss": 0.5717,
|
| 639 |
+
"rewards/accuracies": 0.699999988079071,
|
| 640 |
+
"rewards/chosen": -1.3450363874435425,
|
| 641 |
+
"rewards/margins": 0.736282467842102,
|
| 642 |
+
"rewards/rejected": -2.0813188552856445,
|
| 643 |
+
"step": 420
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 0.42,
|
| 647 |
+
"learning_rate": 3.5725275758125564e-06,
|
| 648 |
+
"logits/chosen": -2.2823150157928467,
|
| 649 |
+
"logits/rejected": -2.1387839317321777,
|
| 650 |
+
"logps/chosen": -135.34524536132812,
|
| 651 |
+
"logps/rejected": -134.94952392578125,
|
| 652 |
+
"loss": 0.4512,
|
| 653 |
+
"rewards/accuracies": 0.824999988079071,
|
| 654 |
+
"rewards/chosen": -1.2761428356170654,
|
| 655 |
+
"rewards/margins": 1.0124304294586182,
|
| 656 |
+
"rewards/rejected": -2.2885732650756836,
|
| 657 |
+
"step": 430
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.43,
|
| 661 |
+
"learning_rate": 3.494289060578478e-06,
|
| 662 |
+
"logits/chosen": -2.3979570865631104,
|
| 663 |
+
"logits/rejected": -2.3127408027648926,
|
| 664 |
+
"logps/chosen": -148.89285278320312,
|
| 665 |
+
"logps/rejected": -157.427978515625,
|
| 666 |
+
"loss": 0.5081,
|
| 667 |
+
"rewards/accuracies": 0.6875,
|
| 668 |
+
"rewards/chosen": -1.634016990661621,
|
| 669 |
+
"rewards/margins": 0.8799861669540405,
|
| 670 |
+
"rewards/rejected": -2.514003276824951,
|
| 671 |
+
"step": 440
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 0.44,
|
| 675 |
+
"learning_rate": 3.414875979798272e-06,
|
| 676 |
+
"logits/chosen": -2.349696636199951,
|
| 677 |
+
"logits/rejected": -2.236372709274292,
|
| 678 |
+
"logps/chosen": -163.39578247070312,
|
| 679 |
+
"logps/rejected": -173.43084716796875,
|
| 680 |
+
"loss": 0.4372,
|
| 681 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 682 |
+
"rewards/chosen": -1.1473162174224854,
|
| 683 |
+
"rewards/margins": 1.2105666399002075,
|
| 684 |
+
"rewards/rejected": -2.3578829765319824,
|
| 685 |
+
"step": 450
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 0.45,
|
| 689 |
+
"learning_rate": 3.3343821450930196e-06,
|
| 690 |
+
"logits/chosen": -2.468815326690674,
|
| 691 |
+
"logits/rejected": -2.319728374481201,
|
| 692 |
+
"logps/chosen": -147.9761199951172,
|
| 693 |
+
"logps/rejected": -140.23440551757812,
|
| 694 |
+
"loss": 0.4893,
|
| 695 |
+
"rewards/accuracies": 0.75,
|
| 696 |
+
"rewards/chosen": -1.0942740440368652,
|
| 697 |
+
"rewards/margins": 1.0382341146469116,
|
| 698 |
+
"rewards/rejected": -2.132507801055908,
|
| 699 |
+
"step": 460
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.46,
|
| 703 |
+
"learning_rate": 3.252902644791325e-06,
|
| 704 |
+
"logits/chosen": -2.2242140769958496,
|
| 705 |
+
"logits/rejected": -2.078382968902588,
|
| 706 |
+
"logps/chosen": -138.49478149414062,
|
| 707 |
+
"logps/rejected": -156.64273071289062,
|
| 708 |
+
"loss": 0.4146,
|
| 709 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 710 |
+
"rewards/chosen": -1.0762813091278076,
|
| 711 |
+
"rewards/margins": 1.254016399383545,
|
| 712 |
+
"rewards/rejected": -2.3302977085113525,
|
| 713 |
+
"step": 470
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 0.47,
|
| 717 |
+
"learning_rate": 3.170533731600339e-06,
|
| 718 |
+
"logits/chosen": -2.334146022796631,
|
| 719 |
+
"logits/rejected": -2.1843693256378174,
|
| 720 |
+
"logps/chosen": -145.91741943359375,
|
| 721 |
+
"logps/rejected": -150.4651336669922,
|
| 722 |
+
"loss": 0.5102,
|
| 723 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 724 |
+
"rewards/chosen": -1.1809916496276855,
|
| 725 |
+
"rewards/margins": 1.0365092754364014,
|
| 726 |
+
"rewards/rejected": -2.217501163482666,
|
| 727 |
+
"step": 480
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 0.48,
|
| 731 |
+
"learning_rate": 3.0873727089012816e-06,
|
| 732 |
+
"logits/chosen": -2.4750094413757324,
|
| 733 |
+
"logits/rejected": -2.3860108852386475,
|
| 734 |
+
"logps/chosen": -168.5699005126953,
|
| 735 |
+
"logps/rejected": -172.41029357910156,
|
| 736 |
+
"loss": 0.522,
|
| 737 |
+
"rewards/accuracies": 0.737500011920929,
|
| 738 |
+
"rewards/chosen": -1.693511962890625,
|
| 739 |
+
"rewards/margins": 1.0143020153045654,
|
| 740 |
+
"rewards/rejected": -2.7078135013580322,
|
| 741 |
+
"step": 490
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 0.49,
|
| 745 |
+
"learning_rate": 3.0035178158038026e-06,
|
| 746 |
+
"logits/chosen": -2.261049747467041,
|
| 747 |
+
"logits/rejected": -2.0733211040496826,
|
| 748 |
+
"logps/chosen": -178.0288543701172,
|
| 749 |
+
"logps/rejected": -181.95970153808594,
|
| 750 |
+
"loss": 0.4784,
|
| 751 |
+
"rewards/accuracies": 0.75,
|
| 752 |
+
"rewards/chosen": -1.8506263494491577,
|
| 753 |
+
"rewards/margins": 1.296245813369751,
|
| 754 |
+
"rewards/rejected": -3.146872043609619,
|
| 755 |
+
"step": 500
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 0.5,
|
| 759 |
+
"learning_rate": 2.919068111094937e-06,
|
| 760 |
+
"logits/chosen": -2.3573222160339355,
|
| 761 |
+
"logits/rejected": -2.2329063415527344,
|
| 762 |
+
"logps/chosen": -169.11073303222656,
|
| 763 |
+
"logps/rejected": -186.6231231689453,
|
| 764 |
+
"loss": 0.6133,
|
| 765 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 766 |
+
"rewards/chosen": -2.2830090522766113,
|
| 767 |
+
"rewards/margins": 1.0279150009155273,
|
| 768 |
+
"rewards/rejected": -3.3109240531921387,
|
| 769 |
+
"step": 510
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 0.51,
|
| 773 |
+
"learning_rate": 2.8341233562197895e-06,
|
| 774 |
+
"logits/chosen": -2.3493080139160156,
|
| 775 |
+
"logits/rejected": -2.2869277000427246,
|
| 776 |
+
"logps/chosen": -156.89324951171875,
|
| 777 |
+
"logps/rejected": -172.35809326171875,
|
| 778 |
+
"loss": 0.4775,
|
| 779 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 780 |
+
"rewards/chosen": -1.8899619579315186,
|
| 781 |
+
"rewards/margins": 1.054958701133728,
|
| 782 |
+
"rewards/rejected": -2.944920778274536,
|
| 783 |
+
"step": 520
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.52,
|
| 787 |
+
"learning_rate": 2.7487838974321352e-06,
|
| 788 |
+
"logits/chosen": -2.294079065322876,
|
| 789 |
+
"logits/rejected": -2.1721999645233154,
|
| 790 |
+
"logps/chosen": -166.351806640625,
|
| 791 |
+
"logps/rejected": -181.2619171142578,
|
| 792 |
+
"loss": 0.501,
|
| 793 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 794 |
+
"rewards/chosen": -2.059480905532837,
|
| 795 |
+
"rewards/margins": 1.2059848308563232,
|
| 796 |
+
"rewards/rejected": -3.265465259552002,
|
| 797 |
+
"step": 530
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 0.53,
|
| 801 |
+
"learning_rate": 2.6631505472541997e-06,
|
| 802 |
+
"logits/chosen": -2.284674644470215,
|
| 803 |
+
"logits/rejected": -2.151689291000366,
|
| 804 |
+
"logps/chosen": -170.77108764648438,
|
| 805 |
+
"logps/rejected": -185.42703247070312,
|
| 806 |
+
"loss": 0.4457,
|
| 807 |
+
"rewards/accuracies": 0.737500011920929,
|
| 808 |
+
"rewards/chosen": -1.8532861471176147,
|
| 809 |
+
"rewards/margins": 1.2717969417572021,
|
| 810 |
+
"rewards/rejected": -3.1250832080841064,
|
| 811 |
+
"step": 540
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 0.54,
|
| 815 |
+
"learning_rate": 2.5773244653856173e-06,
|
| 816 |
+
"logits/chosen": -2.262991428375244,
|
| 817 |
+
"logits/rejected": -2.140122652053833,
|
| 818 |
+
"logps/chosen": -171.88180541992188,
|
| 819 |
+
"logps/rejected": -183.91680908203125,
|
| 820 |
+
"loss": 0.4782,
|
| 821 |
+
"rewards/accuracies": 0.762499988079071,
|
| 822 |
+
"rewards/chosen": -1.8343322277069092,
|
| 823 |
+
"rewards/margins": 1.209377408027649,
|
| 824 |
+
"rewards/rejected": -3.0437099933624268,
|
| 825 |
+
"step": 550
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.55,
|
| 829 |
+
"learning_rate": 2.4914070392022717e-06,
|
| 830 |
+
"logits/chosen": -2.287393093109131,
|
| 831 |
+
"logits/rejected": -2.1789438724517822,
|
| 832 |
+
"logps/chosen": -167.02993774414062,
|
| 833 |
+
"logps/rejected": -178.21737670898438,
|
| 834 |
+
"loss": 0.4847,
|
| 835 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 836 |
+
"rewards/chosen": -1.4670352935791016,
|
| 837 |
+
"rewards/margins": 1.1378867626190186,
|
| 838 |
+
"rewards/rejected": -2.60492205619812,
|
| 839 |
+
"step": 560
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 0.56,
|
| 843 |
+
"learning_rate": 2.4054997639861778e-06,
|
| 844 |
+
"logits/chosen": -2.2121901512145996,
|
| 845 |
+
"logits/rejected": -2.039933443069458,
|
| 846 |
+
"logps/chosen": -164.4762725830078,
|
| 847 |
+
"logps/rejected": -171.06558227539062,
|
| 848 |
+
"loss": 0.4581,
|
| 849 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 850 |
+
"rewards/chosen": -1.45037043094635,
|
| 851 |
+
"rewards/margins": 1.332580327987671,
|
| 852 |
+
"rewards/rejected": -2.7829508781433105,
|
| 853 |
+
"step": 570
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 0.57,
|
| 857 |
+
"learning_rate": 2.3197041230278905e-06,
|
| 858 |
+
"logits/chosen": -2.336905002593994,
|
| 859 |
+
"logits/rejected": -2.2310211658477783,
|
| 860 |
+
"logps/chosen": -159.6435089111328,
|
| 861 |
+
"logps/rejected": -185.4768829345703,
|
| 862 |
+
"loss": 0.4182,
|
| 863 |
+
"rewards/accuracies": 0.824999988079071,
|
| 864 |
+
"rewards/chosen": -1.4801210165023804,
|
| 865 |
+
"rewards/margins": 1.5004690885543823,
|
| 866 |
+
"rewards/rejected": -2.980590343475342,
|
| 867 |
+
"step": 580
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.58,
|
| 871 |
+
"learning_rate": 2.234121467743082e-06,
|
| 872 |
+
"logits/chosen": -2.356192111968994,
|
| 873 |
+
"logits/rejected": -2.269792079925537,
|
| 874 |
+
"logps/chosen": -152.16677856445312,
|
| 875 |
+
"logps/rejected": -166.55563354492188,
|
| 876 |
+
"loss": 0.545,
|
| 877 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 878 |
+
"rewards/chosen": -1.5818209648132324,
|
| 879 |
+
"rewards/margins": 1.1744325160980225,
|
| 880 |
+
"rewards/rejected": -2.756253719329834,
|
| 881 |
+
"step": 590
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 0.59,
|
| 885 |
+
"learning_rate": 2.148852897944905e-06,
|
| 886 |
+
"logits/chosen": -2.4117207527160645,
|
| 887 |
+
"logits/rejected": -2.2672557830810547,
|
| 888 |
+
"logps/chosen": -149.95504760742188,
|
| 889 |
+
"logps/rejected": -165.35263061523438,
|
| 890 |
+
"loss": 0.4388,
|
| 891 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 892 |
+
"rewards/chosen": -1.488303303718567,
|
| 893 |
+
"rewards/margins": 1.227616786956787,
|
| 894 |
+
"rewards/rejected": -2.7159199714660645,
|
| 895 |
+
"step": 600
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 0.59,
|
| 899 |
+
"eval_logits/chosen": -2.1954238414764404,
|
| 900 |
+
"eval_logits/rejected": -2.1017305850982666,
|
| 901 |
+
"eval_logps/chosen": -161.69036865234375,
|
| 902 |
+
"eval_logps/rejected": -173.2707977294922,
|
| 903 |
+
"eval_loss": 0.5324557423591614,
|
| 904 |
+
"eval_rewards/accuracies": 0.7122641801834106,
|
| 905 |
+
"eval_rewards/chosen": -1.8012140989303589,
|
| 906 |
+
"eval_rewards/margins": 1.091520071029663,
|
| 907 |
+
"eval_rewards/rejected": -2.8927345275878906,
|
| 908 |
+
"eval_runtime": 427.5006,
|
| 909 |
+
"eval_samples_per_second": 0.978,
|
| 910 |
+
"eval_steps_per_second": 0.124,
|
| 911 |
+
"step": 600
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 0.6,
|
| 915 |
+
"learning_rate": 2.063999142413574e-06,
|
| 916 |
+
"logits/chosen": -2.3014285564422607,
|
| 917 |
+
"logits/rejected": -2.1665964126586914,
|
| 918 |
+
"logps/chosen": -164.73809814453125,
|
| 919 |
+
"logps/rejected": -163.736083984375,
|
| 920 |
+
"loss": 0.4451,
|
| 921 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 922 |
+
"rewards/chosen": -1.6745765209197998,
|
| 923 |
+
"rewards/margins": 1.155146598815918,
|
| 924 |
+
"rewards/rejected": -2.8297228813171387,
|
| 925 |
+
"step": 610
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 0.61,
|
| 929 |
+
"learning_rate": 1.9796604399042547e-06,
|
| 930 |
+
"logits/chosen": -2.375365972518921,
|
| 931 |
+
"logits/rejected": -2.2335057258605957,
|
| 932 |
+
"logps/chosen": -170.6911163330078,
|
| 933 |
+
"logps/rejected": -186.2073974609375,
|
| 934 |
+
"loss": 0.4407,
|
| 935 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 936 |
+
"rewards/chosen": -1.9759544134140015,
|
| 937 |
+
"rewards/margins": 1.4642221927642822,
|
| 938 |
+
"rewards/rejected": -3.440176486968994,
|
| 939 |
+
"step": 620
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.62,
|
| 943 |
+
"learning_rate": 1.8959364207338216e-06,
|
| 944 |
+
"logits/chosen": -2.3572235107421875,
|
| 945 |
+
"logits/rejected": -2.1937854290008545,
|
| 946 |
+
"logps/chosen": -148.90460205078125,
|
| 947 |
+
"logps/rejected": -174.04818725585938,
|
| 948 |
+
"loss": 0.5111,
|
| 949 |
+
"rewards/accuracies": 0.75,
|
| 950 |
+
"rewards/chosen": -1.8767086267471313,
|
| 951 |
+
"rewards/margins": 1.4870884418487549,
|
| 952 |
+
"rewards/rejected": -3.363797426223755,
|
| 953 |
+
"step": 630
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 0.63,
|
| 957 |
+
"learning_rate": 1.8129259890863825e-06,
|
| 958 |
+
"logits/chosen": -2.3194022178649902,
|
| 959 |
+
"logits/rejected": -2.2175240516662598,
|
| 960 |
+
"logps/chosen": -165.7982177734375,
|
| 961 |
+
"logps/rejected": -185.64183044433594,
|
| 962 |
+
"loss": 0.5652,
|
| 963 |
+
"rewards/accuracies": 0.6875,
|
| 964 |
+
"rewards/chosen": -1.662651777267456,
|
| 965 |
+
"rewards/margins": 1.1808096170425415,
|
| 966 |
+
"rewards/rejected": -2.843461513519287,
|
| 967 |
+
"step": 640
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 0.64,
|
| 971 |
+
"learning_rate": 1.7307272061765738e-06,
|
| 972 |
+
"logits/chosen": -2.355379819869995,
|
| 973 |
+
"logits/rejected": -2.256904125213623,
|
| 974 |
+
"logps/chosen": -160.78778076171875,
|
| 975 |
+
"logps/rejected": -178.40591430664062,
|
| 976 |
+
"loss": 0.489,
|
| 977 |
+
"rewards/accuracies": 0.8125,
|
| 978 |
+
"rewards/chosen": -1.5761786699295044,
|
| 979 |
+
"rewards/margins": 1.2927827835083008,
|
| 980 |
+
"rewards/rejected": -2.8689613342285156,
|
| 981 |
+
"step": 650
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 0.65,
|
| 985 |
+
"learning_rate": 1.649437174408685e-06,
|
| 986 |
+
"logits/chosen": -2.2938830852508545,
|
| 987 |
+
"logits/rejected": -2.162132740020752,
|
| 988 |
+
"logps/chosen": -150.61842346191406,
|
| 989 |
+
"logps/rejected": -170.839111328125,
|
| 990 |
+
"loss": 0.5404,
|
| 991 |
+
"rewards/accuracies": 0.737500011920929,
|
| 992 |
+
"rewards/chosen": -1.1881531476974487,
|
| 993 |
+
"rewards/margins": 1.283790111541748,
|
| 994 |
+
"rewards/rejected": -2.4719433784484863,
|
| 995 |
+
"step": 660
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 0.66,
|
| 999 |
+
"learning_rate": 1.569151922668422e-06,
|
| 1000 |
+
"logits/chosen": -2.3966734409332275,
|
| 1001 |
+
"logits/rejected": -2.287233829498291,
|
| 1002 |
+
"logps/chosen": -138.367431640625,
|
| 1003 |
+
"logps/rejected": -159.68118286132812,
|
| 1004 |
+
"loss": 0.5196,
|
| 1005 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1006 |
+
"rewards/chosen": -1.2400791645050049,
|
| 1007 |
+
"rewards/margins": 1.232775330543518,
|
| 1008 |
+
"rewards/rejected": -2.4728543758392334,
|
| 1009 |
+
"step": 670
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 0.67,
|
| 1013 |
+
"learning_rate": 1.4899662928828428e-06,
|
| 1014 |
+
"logits/chosen": -2.333634853363037,
|
| 1015 |
+
"logits/rejected": -2.157811403274536,
|
| 1016 |
+
"logps/chosen": -127.2032699584961,
|
| 1017 |
+
"logps/rejected": -149.32907104492188,
|
| 1018 |
+
"loss": 0.4387,
|
| 1019 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1020 |
+
"rewards/chosen": -1.0127990245819092,
|
| 1021 |
+
"rewards/margins": 1.4315626621246338,
|
| 1022 |
+
"rewards/rejected": -2.444361686706543,
|
| 1023 |
+
"step": 680
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 0.68,
|
| 1027 |
+
"learning_rate": 1.4119738279824507e-06,
|
| 1028 |
+
"logits/chosen": -2.272979259490967,
|
| 1029 |
+
"logits/rejected": -2.127248764038086,
|
| 1030 |
+
"logps/chosen": -131.46400451660156,
|
| 1031 |
+
"logps/rejected": -144.866455078125,
|
| 1032 |
+
"loss": 0.5303,
|
| 1033 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1034 |
+
"rewards/chosen": -1.0165386199951172,
|
| 1035 |
+
"rewards/margins": 1.0676348209381104,
|
| 1036 |
+
"rewards/rejected": -2.0841732025146484,
|
| 1037 |
+
"step": 690
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 0.69,
|
| 1041 |
+
"learning_rate": 1.3352666613978152e-06,
|
| 1042 |
+
"logits/chosen": -2.192830801010132,
|
| 1043 |
+
"logits/rejected": -2.111144542694092,
|
| 1044 |
+
"logps/chosen": -134.23226928710938,
|
| 1045 |
+
"logps/rejected": -143.62905883789062,
|
| 1046 |
+
"loss": 0.5064,
|
| 1047 |
+
"rewards/accuracies": 0.75,
|
| 1048 |
+
"rewards/chosen": -1.0449942350387573,
|
| 1049 |
+
"rewards/margins": 1.0384838581085205,
|
| 1050 |
+
"rewards/rejected": -2.0834782123565674,
|
| 1051 |
+
"step": 700
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 0.7,
|
| 1055 |
+
"learning_rate": 1.2599354082212523e-06,
|
| 1056 |
+
"logits/chosen": -2.2571260929107666,
|
| 1057 |
+
"logits/rejected": -2.1350224018096924,
|
| 1058 |
+
"logps/chosen": -129.79287719726562,
|
| 1059 |
+
"logps/rejected": -138.2635498046875,
|
| 1060 |
+
"loss": 0.5149,
|
| 1061 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1062 |
+
"rewards/chosen": -0.9037138223648071,
|
| 1063 |
+
"rewards/margins": 0.9939008951187134,
|
| 1064 |
+
"rewards/rejected": -1.8976147174835205,
|
| 1065 |
+
"step": 710
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.71,
|
| 1069 |
+
"learning_rate": 1.186069058162127e-06,
|
| 1070 |
+
"logits/chosen": -2.3779895305633545,
|
| 1071 |
+
"logits/rejected": -2.2798473834991455,
|
| 1072 |
+
"logps/chosen": -136.55984497070312,
|
| 1073 |
+
"logps/rejected": -145.81918334960938,
|
| 1074 |
+
"loss": 0.5318,
|
| 1075 |
+
"rewards/accuracies": 0.75,
|
| 1076 |
+
"rewards/chosen": -1.2205983400344849,
|
| 1077 |
+
"rewards/margins": 0.9511237144470215,
|
| 1078 |
+
"rewards/rejected": -2.171722173690796,
|
| 1079 |
+
"step": 720
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 0.72,
|
| 1083 |
+
"learning_rate": 1.113754870422254e-06,
|
| 1084 |
+
"logits/chosen": -2.4235808849334717,
|
| 1085 |
+
"logits/rejected": -2.3193821907043457,
|
| 1086 |
+
"logps/chosen": -142.02517700195312,
|
| 1087 |
+
"logps/rejected": -161.961669921875,
|
| 1088 |
+
"loss": 0.5042,
|
| 1089 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1090 |
+
"rewards/chosen": -1.2923016548156738,
|
| 1091 |
+
"rewards/margins": 0.9127277135848999,
|
| 1092 |
+
"rewards/rejected": -2.205029249191284,
|
| 1093 |
+
"step": 730
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 0.73,
|
| 1097 |
+
"learning_rate": 1.0430782706155545e-06,
|
| 1098 |
+
"logits/chosen": -2.435793399810791,
|
| 1099 |
+
"logits/rejected": -2.296560764312744,
|
| 1100 |
+
"logps/chosen": -145.57290649414062,
|
| 1101 |
+
"logps/rejected": -156.23214721679688,
|
| 1102 |
+
"loss": 0.4346,
|
| 1103 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1104 |
+
"rewards/chosen": -1.0421411991119385,
|
| 1105 |
+
"rewards/margins": 1.3148539066314697,
|
| 1106 |
+
"rewards/rejected": -2.356995105743408,
|
| 1107 |
+
"step": 740
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 0.74,
|
| 1111 |
+
"learning_rate": 9.741227498537615e-07,
|
| 1112 |
+
"logits/chosen": -2.4774973392486572,
|
| 1113 |
+
"logits/rejected": -2.3496251106262207,
|
| 1114 |
+
"logps/chosen": -133.50381469726562,
|
| 1115 |
+
"logps/rejected": -148.293701171875,
|
| 1116 |
+
"loss": 0.4295,
|
| 1117 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1118 |
+
"rewards/chosen": -1.2879221439361572,
|
| 1119 |
+
"rewards/margins": 1.1688247919082642,
|
| 1120 |
+
"rewards/rejected": -2.45674729347229,
|
| 1121 |
+
"step": 750
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 0.75,
|
| 1125 |
+
"learning_rate": 9.069697661173668e-07,
|
| 1126 |
+
"logits/chosen": -2.3304691314697266,
|
| 1127 |
+
"logits/rejected": -2.2422666549682617,
|
| 1128 |
+
"logps/chosen": -143.60433959960938,
|
| 1129 |
+
"logps/rejected": -161.01876831054688,
|
| 1130 |
+
"loss": 0.6187,
|
| 1131 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1132 |
+
"rewards/chosen": -1.4187496900558472,
|
| 1133 |
+
"rewards/margins": 0.990622878074646,
|
| 1134 |
+
"rewards/rejected": -2.4093728065490723,
|
| 1135 |
+
"step": 760
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 0.76,
|
| 1139 |
+
"learning_rate": 8.416986480283434e-07,
|
| 1140 |
+
"logits/chosen": -2.2641918659210205,
|
| 1141 |
+
"logits/rejected": -2.1294054985046387,
|
| 1142 |
+
"logps/chosen": -150.9503173828125,
|
| 1143 |
+
"logps/rejected": -169.94259643554688,
|
| 1144 |
+
"loss": 0.5414,
|
| 1145 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1146 |
+
"rewards/chosen": -1.6702228784561157,
|
| 1147 |
+
"rewards/margins": 1.2004982233047485,
|
| 1148 |
+
"rewards/rejected": -2.8707211017608643,
|
| 1149 |
+
"step": 770
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.77,
|
| 1153 |
+
"learning_rate": 7.783865011382876e-07,
|
| 1154 |
+
"logits/chosen": -2.318986415863037,
|
| 1155 |
+
"logits/rejected": -2.189507246017456,
|
| 1156 |
+
"logps/chosen": -157.97384643554688,
|
| 1157 |
+
"logps/rejected": -164.84359741210938,
|
| 1158 |
+
"loss": 0.4953,
|
| 1159 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1160 |
+
"rewards/chosen": -1.3389387130737305,
|
| 1161 |
+
"rewards/margins": 1.2802064418792725,
|
| 1162 |
+
"rewards/rejected": -2.619144916534424,
|
| 1163 |
+
"step": 780
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 0.78,
|
| 1167 |
+
"learning_rate": 7.171081168427205e-07,
|
| 1168 |
+
"logits/chosen": -2.3565828800201416,
|
| 1169 |
+
"logits/rejected": -2.193580150604248,
|
| 1170 |
+
"logps/chosen": -169.49929809570312,
|
| 1171 |
+
"logps/rejected": -177.23731994628906,
|
| 1172 |
+
"loss": 0.5052,
|
| 1173 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1174 |
+
"rewards/chosen": -1.3413608074188232,
|
| 1175 |
+
"rewards/margins": 1.271929144859314,
|
| 1176 |
+
"rewards/rejected": -2.6132900714874268,
|
| 1177 |
+
"step": 790
|
| 1178 |
+
},
|
| 1179 |
+
{
|
| 1180 |
+
"epoch": 0.79,
|
| 1181 |
+
"learning_rate": 6.579358840291064e-07,
|
| 1182 |
+
"logits/chosen": -2.3053181171417236,
|
| 1183 |
+
"logits/rejected": -2.205723762512207,
|
| 1184 |
+
"logps/chosen": -155.66004943847656,
|
| 1185 |
+
"logps/rejected": -173.5889434814453,
|
| 1186 |
+
"loss": 0.6137,
|
| 1187 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1188 |
+
"rewards/chosen": -1.5923388004302979,
|
| 1189 |
+
"rewards/margins": 0.8797184228897095,
|
| 1190 |
+
"rewards/rejected": -2.4720568656921387,
|
| 1191 |
+
"step": 800
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 0.79,
|
| 1195 |
+
"eval_logits/chosen": -2.19348406791687,
|
| 1196 |
+
"eval_logits/rejected": -2.0962369441986084,
|
| 1197 |
+
"eval_logps/chosen": -149.9387664794922,
|
| 1198 |
+
"eval_logps/rejected": -160.84133911132812,
|
| 1199 |
+
"eval_loss": 0.5198491215705872,
|
| 1200 |
+
"eval_rewards/accuracies": 0.7382075190544128,
|
| 1201 |
+
"eval_rewards/chosen": -1.448665976524353,
|
| 1202 |
+
"eval_rewards/margins": 1.0711843967437744,
|
| 1203 |
+
"eval_rewards/rejected": -2.519850254058838,
|
| 1204 |
+
"eval_runtime": 426.9839,
|
| 1205 |
+
"eval_samples_per_second": 0.979,
|
| 1206 |
+
"eval_steps_per_second": 0.124,
|
| 1207 |
+
"step": 800
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.8,
|
| 1211 |
+
"learning_rate": 6.00939703563006e-07,
|
| 1212 |
+
"logits/chosen": -2.3648667335510254,
|
| 1213 |
+
"logits/rejected": -2.2830593585968018,
|
| 1214 |
+
"logps/chosen": -136.9246826171875,
|
| 1215 |
+
"logps/rejected": -156.57559204101562,
|
| 1216 |
+
"loss": 0.5648,
|
| 1217 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1218 |
+
"rewards/chosen": -1.4630320072174072,
|
| 1219 |
+
"rewards/margins": 0.9899848699569702,
|
| 1220 |
+
"rewards/rejected": -2.453016757965088,
|
| 1221 |
+
"step": 810
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.81,
|
| 1225 |
+
"learning_rate": 5.461869057133412e-07,
|
| 1226 |
+
"logits/chosen": -2.408620595932007,
|
| 1227 |
+
"logits/rejected": -2.319190502166748,
|
| 1228 |
+
"logps/chosen": -145.22872924804688,
|
| 1229 |
+
"logps/rejected": -167.64926147460938,
|
| 1230 |
+
"loss": 0.487,
|
| 1231 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1232 |
+
"rewards/chosen": -1.306095838546753,
|
| 1233 |
+
"rewards/margins": 1.1592603921890259,
|
| 1234 |
+
"rewards/rejected": -2.4653563499450684,
|
| 1235 |
+
"step": 820
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 0.82,
|
| 1239 |
+
"learning_rate": 4.937421706143497e-07,
|
| 1240 |
+
"logits/chosen": -2.290102481842041,
|
| 1241 |
+
"logits/rejected": -2.158676862716675,
|
| 1242 |
+
"logps/chosen": -152.98867797851562,
|
| 1243 |
+
"logps/rejected": -159.16566467285156,
|
| 1244 |
+
"loss": 0.5218,
|
| 1245 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1246 |
+
"rewards/chosen": -1.3972251415252686,
|
| 1247 |
+
"rewards/margins": 1.101043462753296,
|
| 1248 |
+
"rewards/rejected": -2.4982683658599854,
|
| 1249 |
+
"step": 830
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.83,
|
| 1253 |
+
"learning_rate": 4.43667451858166e-07,
|
| 1254 |
+
"logits/chosen": -2.2902016639709473,
|
| 1255 |
+
"logits/rejected": -2.131404399871826,
|
| 1256 |
+
"logps/chosen": -134.6024627685547,
|
| 1257 |
+
"logps/rejected": -151.9069061279297,
|
| 1258 |
+
"loss": 0.4313,
|
| 1259 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1260 |
+
"rewards/chosen": -1.2463924884796143,
|
| 1261 |
+
"rewards/margins": 1.268364667892456,
|
| 1262 |
+
"rewards/rejected": -2.514756917953491,
|
| 1263 |
+
"step": 840
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.84,
|
| 1267 |
+
"learning_rate": 3.9602190330830484e-07,
|
| 1268 |
+
"logits/chosen": -2.2570953369140625,
|
| 1269 |
+
"logits/rejected": -2.157632827758789,
|
| 1270 |
+
"logps/chosen": -141.2190704345703,
|
| 1271 |
+
"logps/rejected": -162.86306762695312,
|
| 1272 |
+
"loss": 0.4876,
|
| 1273 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1274 |
+
"rewards/chosen": -1.3870246410369873,
|
| 1275 |
+
"rewards/margins": 1.1176416873931885,
|
| 1276 |
+
"rewards/rejected": -2.504666328430176,
|
| 1277 |
+
"step": 850
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 0.85,
|
| 1281 |
+
"learning_rate": 3.5086180922049295e-07,
|
| 1282 |
+
"logits/chosen": -2.394253969192505,
|
| 1283 |
+
"logits/rejected": -2.2592263221740723,
|
| 1284 |
+
"logps/chosen": -165.12005615234375,
|
| 1285 |
+
"logps/rejected": -166.54446411132812,
|
| 1286 |
+
"loss": 0.4366,
|
| 1287 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1288 |
+
"rewards/chosen": -1.3376173973083496,
|
| 1289 |
+
"rewards/margins": 1.2365379333496094,
|
| 1290 |
+
"rewards/rejected": -2.574155330657959,
|
| 1291 |
+
"step": 860
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.86,
|
| 1295 |
+
"learning_rate": 3.0824051775340895e-07,
|
| 1296 |
+
"logits/chosen": -2.424966335296631,
|
| 1297 |
+
"logits/rejected": -2.3199541568756104,
|
| 1298 |
+
"logps/chosen": -124.06752014160156,
|
| 1299 |
+
"logps/rejected": -140.36489868164062,
|
| 1300 |
+
"loss": 0.5696,
|
| 1301 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1302 |
+
"rewards/chosen": -1.2786959409713745,
|
| 1303 |
+
"rewards/margins": 0.8720223307609558,
|
| 1304 |
+
"rewards/rejected": -2.1507184505462646,
|
| 1305 |
+
"step": 870
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.87,
|
| 1309 |
+
"learning_rate": 2.6820837794786336e-07,
|
| 1310 |
+
"logits/chosen": -2.2567453384399414,
|
| 1311 |
+
"logits/rejected": -2.1820857524871826,
|
| 1312 |
+
"logps/chosen": -145.2144317626953,
|
| 1313 |
+
"logps/rejected": -155.57261657714844,
|
| 1314 |
+
"loss": 0.6788,
|
| 1315 |
+
"rewards/accuracies": 0.574999988079071,
|
| 1316 |
+
"rewards/chosen": -1.4533874988555908,
|
| 1317 |
+
"rewards/margins": 0.4883599281311035,
|
| 1318 |
+
"rewards/rejected": -1.9417474269866943,
|
| 1319 |
+
"step": 880
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.88,
|
| 1323 |
+
"learning_rate": 2.3081268024887694e-07,
|
| 1324 |
+
"logits/chosen": -2.2515053749084473,
|
| 1325 |
+
"logits/rejected": -2.0878632068634033,
|
| 1326 |
+
"logps/chosen": -141.62442016601562,
|
| 1327 |
+
"logps/rejected": -151.8519287109375,
|
| 1328 |
+
"loss": 0.4185,
|
| 1329 |
+
"rewards/accuracies": 0.8125,
|
| 1330 |
+
"rewards/chosen": -1.220234990119934,
|
| 1331 |
+
"rewards/margins": 1.2837200164794922,
|
| 1332 |
+
"rewards/rejected": -2.5039544105529785,
|
| 1333 |
+
"step": 890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 0.89,
|
| 1337 |
+
"learning_rate": 1.9609760064091044e-07,
|
| 1338 |
+
"logits/chosen": -2.343606472015381,
|
| 1339 |
+
"logits/rejected": -2.277740955352783,
|
| 1340 |
+
"logps/chosen": -144.82054138183594,
|
| 1341 |
+
"logps/rejected": -142.51901245117188,
|
| 1342 |
+
"loss": 0.5214,
|
| 1343 |
+
"rewards/accuracies": 0.75,
|
| 1344 |
+
"rewards/chosen": -1.2351983785629272,
|
| 1345 |
+
"rewards/margins": 0.8959378004074097,
|
| 1346 |
+
"rewards/rejected": -2.131136417388916,
|
| 1347 |
+
"step": 900
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.9,
|
| 1351 |
+
"learning_rate": 1.6410414846224992e-07,
|
| 1352 |
+
"logits/chosen": -2.256499767303467,
|
| 1353 |
+
"logits/rejected": -2.1499085426330566,
|
| 1354 |
+
"logps/chosen": -139.65145874023438,
|
| 1355 |
+
"logps/rejected": -156.27642822265625,
|
| 1356 |
+
"loss": 0.429,
|
| 1357 |
+
"rewards/accuracies": 0.8125,
|
| 1358 |
+
"rewards/chosen": -1.2983112335205078,
|
| 1359 |
+
"rewards/margins": 1.2496817111968994,
|
| 1360 |
+
"rewards/rejected": -2.547992706298828,
|
| 1361 |
+
"step": 910
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.91,
|
| 1365 |
+
"learning_rate": 1.348701179601819e-07,
|
| 1366 |
+
"logits/chosen": -2.4174904823303223,
|
| 1367 |
+
"logits/rejected": -2.2869112491607666,
|
| 1368 |
+
"logps/chosen": -156.2935028076172,
|
| 1369 |
+
"logps/rejected": -166.49362182617188,
|
| 1370 |
+
"loss": 0.4412,
|
| 1371 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1372 |
+
"rewards/chosen": -1.221897840499878,
|
| 1373 |
+
"rewards/margins": 1.1692349910736084,
|
| 1374 |
+
"rewards/rejected": -2.3911328315734863,
|
| 1375 |
+
"step": 920
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.91,
|
| 1379 |
+
"learning_rate": 1.0843004364420151e-07,
|
| 1380 |
+
"logits/chosen": -2.2345454692840576,
|
| 1381 |
+
"logits/rejected": -2.1238150596618652,
|
| 1382 |
+
"logps/chosen": -157.7463836669922,
|
| 1383 |
+
"logps/rejected": -174.59848022460938,
|
| 1384 |
+
"loss": 0.5407,
|
| 1385 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1386 |
+
"rewards/chosen": -1.309340476989746,
|
| 1387 |
+
"rewards/margins": 0.9167596101760864,
|
| 1388 |
+
"rewards/rejected": -2.226100206375122,
|
| 1389 |
+
"step": 930
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.92,
|
| 1393 |
+
"learning_rate": 8.481515948997931e-08,
|
| 1394 |
+
"logits/chosen": -2.384162425994873,
|
| 1395 |
+
"logits/rejected": -2.2841312885284424,
|
| 1396 |
+
"logps/chosen": -161.34854125976562,
|
| 1397 |
+
"logps/rejected": -156.95523071289062,
|
| 1398 |
+
"loss": 0.6309,
|
| 1399 |
+
"rewards/accuracies": 0.75,
|
| 1400 |
+
"rewards/chosen": -1.4349567890167236,
|
| 1401 |
+
"rewards/margins": 0.7904176115989685,
|
| 1402 |
+
"rewards/rejected": -2.225374460220337,
|
| 1403 |
+
"step": 940
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 0.93,
|
| 1407 |
+
"learning_rate": 6.4053362042297e-08,
|
| 1408 |
+
"logits/chosen": -2.284637928009033,
|
| 1409 |
+
"logits/rejected": -2.121169090270996,
|
| 1410 |
+
"logps/chosen": -147.6913299560547,
|
| 1411 |
+
"logps/rejected": -160.8138885498047,
|
| 1412 |
+
"loss": 0.4361,
|
| 1413 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1414 |
+
"rewards/chosen": -1.3165013790130615,
|
| 1415 |
+
"rewards/margins": 1.2154157161712646,
|
| 1416 |
+
"rewards/rejected": -2.5319173336029053,
|
| 1417 |
+
"step": 950
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 0.94,
|
| 1421 |
+
"learning_rate": 4.616917746052163e-08,
|
| 1422 |
+
"logits/chosen": -2.3731329441070557,
|
| 1423 |
+
"logits/rejected": -2.2411139011383057,
|
| 1424 |
+
"logps/chosen": -147.39578247070312,
|
| 1425 |
+
"logps/rejected": -155.5648651123047,
|
| 1426 |
+
"loss": 0.5548,
|
| 1427 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1428 |
+
"rewards/chosen": -1.4831774234771729,
|
| 1429 |
+
"rewards/margins": 1.0112624168395996,
|
| 1430 |
+
"rewards/rejected": -2.4944396018981934,
|
| 1431 |
+
"step": 960
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.95,
|
| 1435 |
+
"learning_rate": 3.118373254556412e-08,
|
| 1436 |
+
"logits/chosen": -2.4090075492858887,
|
| 1437 |
+
"logits/rejected": -2.2784786224365234,
|
| 1438 |
+
"logps/chosen": -138.59571838378906,
|
| 1439 |
+
"logps/rejected": -139.35055541992188,
|
| 1440 |
+
"loss": 0.4862,
|
| 1441 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1442 |
+
"rewards/chosen": -1.2963029146194458,
|
| 1443 |
+
"rewards/margins": 1.0239927768707275,
|
| 1444 |
+
"rewards/rejected": -2.320295810699463,
|
| 1445 |
+
"step": 970
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 0.96,
|
| 1449 |
+
"learning_rate": 1.9114729782535037e-08,
|
| 1450 |
+
"logits/chosen": -2.432577133178711,
|
| 1451 |
+
"logits/rejected": -2.3220748901367188,
|
| 1452 |
+
"logps/chosen": -136.9718017578125,
|
| 1453 |
+
"logps/rejected": -146.90240478515625,
|
| 1454 |
+
"loss": 0.5037,
|
| 1455 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1456 |
+
"rewards/chosen": -1.215503454208374,
|
| 1457 |
+
"rewards/margins": 0.9537515640258789,
|
| 1458 |
+
"rewards/rejected": -2.169254779815674,
|
| 1459 |
+
"step": 980
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 0.97,
|
| 1463 |
+
"learning_rate": 9.97642642858815e-09,
|
| 1464 |
+
"logits/chosen": -2.30542254447937,
|
| 1465 |
+
"logits/rejected": -2.1695542335510254,
|
| 1466 |
+
"logps/chosen": -152.14138793945312,
|
| 1467 |
+
"logps/rejected": -153.89279174804688,
|
| 1468 |
+
"loss": 0.5001,
|
| 1469 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1470 |
+
"rewards/chosen": -1.3888603448867798,
|
| 1471 |
+
"rewards/margins": 0.9819372296333313,
|
| 1472 |
+
"rewards/rejected": -2.3707973957061768,
|
| 1473 |
+
"step": 990
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 0.98,
|
| 1477 |
+
"learning_rate": 3.779617670651436e-09,
|
| 1478 |
+
"logits/chosen": -2.278373956680298,
|
| 1479 |
+
"logits/rejected": -2.17798113822937,
|
| 1480 |
+
"logps/chosen": -157.8939208984375,
|
| 1481 |
+
"logps/rejected": -168.04055786132812,
|
| 1482 |
+
"loss": 0.5866,
|
| 1483 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1484 |
+
"rewards/chosen": -1.575810194015503,
|
| 1485 |
+
"rewards/margins": 0.995488166809082,
|
| 1486 |
+
"rewards/rejected": -2.571298599243164,
|
| 1487 |
+
"step": 1000
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.98,
|
| 1491 |
+
"eval_logits/chosen": -2.192172050476074,
|
| 1492 |
+
"eval_logits/rejected": -2.095205545425415,
|
| 1493 |
+
"eval_logps/chosen": -147.71951293945312,
|
| 1494 |
+
"eval_logps/rejected": -158.54696655273438,
|
| 1495 |
+
"eval_loss": 0.5194684863090515,
|
| 1496 |
+
"eval_rewards/accuracies": 0.7358490824699402,
|
| 1497 |
+
"eval_rewards/chosen": -1.382088541984558,
|
| 1498 |
+
"eval_rewards/margins": 1.068930983543396,
|
| 1499 |
+
"eval_rewards/rejected": -2.451019525527954,
|
| 1500 |
+
"eval_runtime": 427.3583,
|
| 1501 |
+
"eval_samples_per_second": 0.978,
|
| 1502 |
+
"eval_steps_per_second": 0.124,
|
| 1503 |
+
"step": 1000
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 0.99,
|
| 1507 |
+
"learning_rate": 5.316238729444201e-10,
|
| 1508 |
+
"logits/chosen": -2.2608113288879395,
|
| 1509 |
+
"logits/rejected": -2.101694345474243,
|
| 1510 |
+
"logps/chosen": -152.0623779296875,
|
| 1511 |
+
"logps/rejected": -162.7583770751953,
|
| 1512 |
+
"loss": 0.4821,
|
| 1513 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1514 |
+
"rewards/chosen": -1.5301237106323242,
|
| 1515 |
+
"rewards/margins": 1.2161496877670288,
|
| 1516 |
+
"rewards/rejected": -2.7462732791900635,
|
| 1517 |
+
"step": 1010
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 1.0,
|
| 1521 |
+
"step": 1016,
|
| 1522 |
+
"total_flos": 0.0,
|
| 1523 |
+
"train_loss": 0.5536013367607837,
|
| 1524 |
+
"train_runtime": 17040.3461,
|
| 1525 |
+
"train_samples_per_second": 0.477,
|
| 1526 |
+
"train_steps_per_second": 0.06
|
| 1527 |
+
}
|
| 1528 |
+
],
|
| 1529 |
+
"logging_steps": 10,
|
| 1530 |
+
"max_steps": 1016,
|
| 1531 |
+
"num_input_tokens_seen": 0,
|
| 1532 |
+
"num_train_epochs": 1,
|
| 1533 |
+
"save_steps": 100,
|
| 1534 |
+
"total_flos": 0.0,
|
| 1535 |
+
"train_batch_size": 4,
|
| 1536 |
+
"trial_name": null,
|
| 1537 |
+
"trial_params": null
|
| 1538 |
+
}
|