Model save
Browse files- README.md +87 -0
- adapter_model.safetensors +1 -1
- all_results.json +8 -0
- runs/Jul16_15-00-29_node25/events.out.tfevents.1721109642.node25.1416743.0 +2 -2
- train_results.json +8 -0
- trainer_state.json +2146 -0
README.md
ADDED
|
@@ -0,0 +1,87 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
base_model: openbmb/Eurus-7b-sft
|
| 9 |
+
model-index:
|
| 10 |
+
- name: eurus-dpo-qlora-uf-ours-5e-7
|
| 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 |
+
# eurus-dpo-qlora-uf-ours-5e-7
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [openbmb/Eurus-7b-sft](https://huggingface.co/openbmb/Eurus-7b-sft) on the None dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.8251
|
| 22 |
+
- Rewards/chosen: -2.6509
|
| 23 |
+
- Rewards/rejected: -3.1193
|
| 24 |
+
- Rewards/accuracies: 0.5930
|
| 25 |
+
- Rewards/margins: 0.4684
|
| 26 |
+
- Rewards/margins Max: 3.5001
|
| 27 |
+
- Rewards/margins Min: -2.2741
|
| 28 |
+
- Rewards/margins Std: 1.9114
|
| 29 |
+
- Logps/rejected: -569.4471
|
| 30 |
+
- Logps/chosen: -539.9697
|
| 31 |
+
- Logits/rejected: -1.8277
|
| 32 |
+
- Logits/chosen: -1.9148
|
| 33 |
+
|
| 34 |
+
## Model description
|
| 35 |
+
|
| 36 |
+
More information needed
|
| 37 |
+
|
| 38 |
+
## Intended uses & limitations
|
| 39 |
+
|
| 40 |
+
More information needed
|
| 41 |
+
|
| 42 |
+
## Training and evaluation data
|
| 43 |
+
|
| 44 |
+
More information needed
|
| 45 |
+
|
| 46 |
+
## Training procedure
|
| 47 |
+
|
| 48 |
+
### Training hyperparameters
|
| 49 |
+
|
| 50 |
+
The following hyperparameters were used during training:
|
| 51 |
+
- learning_rate: 5e-07
|
| 52 |
+
- train_batch_size: 4
|
| 53 |
+
- eval_batch_size: 8
|
| 54 |
+
- seed: 42
|
| 55 |
+
- distributed_type: multi-GPU
|
| 56 |
+
- num_devices: 2
|
| 57 |
+
- gradient_accumulation_steps: 2
|
| 58 |
+
- total_train_batch_size: 16
|
| 59 |
+
- total_eval_batch_size: 16
|
| 60 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 61 |
+
- lr_scheduler_type: cosine
|
| 62 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 63 |
+
- num_epochs: 3
|
| 64 |
+
|
| 65 |
+
### Training results
|
| 66 |
+
|
| 67 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Rewards/margins Max | Rewards/margins Min | Rewards/margins Std | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 68 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:-------------------:|:-------------------:|:-------------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 69 |
+
| 0.6787 | 0.28 | 100 | 0.6902 | -0.0196 | -0.0275 | 0.6050 | 0.0078 | 0.0658 | -0.0439 | 0.0352 | -260.2682 | -276.8446 | -2.1835 | -2.3057 |
|
| 70 |
+
| 0.6038 | 0.56 | 200 | 0.6829 | -0.2121 | -0.2562 | 0.5930 | 0.0440 | 0.4186 | -0.2883 | 0.2265 | -283.1364 | -296.0924 | -2.1563 | -2.2736 |
|
| 71 |
+
| 0.4746 | 0.85 | 300 | 0.7105 | -0.7773 | -0.8546 | 0.5660 | 0.0773 | 1.0401 | -0.8434 | 0.6093 | -342.9795 | -352.6140 | -2.0904 | -2.1991 |
|
| 72 |
+
| 0.4288 | 1.13 | 400 | 0.7566 | -1.3505 | -1.4749 | 0.5700 | 0.1245 | 1.6613 | -1.3515 | 0.9884 | -405.0142 | -409.9261 | -2.0237 | -2.1254 |
|
| 73 |
+
| 0.3807 | 1.41 | 500 | 0.7770 | -1.7690 | -1.9759 | 0.5760 | 0.2069 | 2.1466 | -1.6287 | 1.2537 | -455.1077 | -451.7817 | -1.9637 | -2.0584 |
|
| 74 |
+
| 0.3449 | 1.69 | 600 | 0.8093 | -2.3053 | -2.6236 | 0.5730 | 0.3183 | 2.7910 | -1.9845 | 1.5908 | -519.8788 | -505.4114 | -1.8829 | -1.9707 |
|
| 75 |
+
| 0.3253 | 1.97 | 700 | 0.8022 | -2.3688 | -2.7622 | 0.5900 | 0.3934 | 3.0600 | -2.0479 | 1.6969 | -533.7401 | -511.7566 | -1.8637 | -1.9524 |
|
| 76 |
+
| 0.2445 | 2.25 | 800 | 0.8262 | -2.6179 | -3.0584 | 0.5880 | 0.4405 | 3.3852 | -2.2378 | 1.8658 | -563.3621 | -536.6691 | -1.8329 | -1.9194 |
|
| 77 |
+
| 0.3015 | 2.54 | 900 | 0.8293 | -2.6774 | -3.1416 | 0.5930 | 0.4642 | 3.5043 | -2.2912 | 1.9184 | -571.6796 | -542.6185 | -1.8281 | -1.9147 |
|
| 78 |
+
| 0.2725 | 2.82 | 1000 | 0.8251 | -2.6509 | -3.1193 | 0.5930 | 0.4684 | 3.5001 | -2.2741 | 1.9114 | -569.4471 | -539.9697 | -1.8277 | -1.9148 |
|
| 79 |
+
|
| 80 |
+
|
| 81 |
+
### Framework versions
|
| 82 |
+
|
| 83 |
+
- PEFT 0.7.1
|
| 84 |
+
- Transformers 4.39.0.dev0
|
| 85 |
+
- Pytorch 2.1.2+cu121
|
| 86 |
+
- Datasets 2.14.6
|
| 87 |
+
- Tokenizers 0.15.2
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 671150064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:de3069743f4f68e5e156c56059cf3ab89aa69b467d1185ff1a06b4c4fb5e5f2f
|
| 3 |
size 671150064
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"train_loss": 0.4186206150502666,
|
| 4 |
+
"train_runtime": 20980.8825,
|
| 5 |
+
"train_samples": 5678,
|
| 6 |
+
"train_samples_per_second": 0.812,
|
| 7 |
+
"train_steps_per_second": 0.051
|
| 8 |
+
}
|
runs/Jul16_15-00-29_node25/events.out.tfevents.1721109642.node25.1416743.0
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 sha256:2218acff401ab1091f48ff643166113dfad1733afd8009e8b374a7fb7049fbc2
|
| 3 |
+
size 108890
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"train_loss": 0.4186206150502666,
|
| 4 |
+
"train_runtime": 20980.8825,
|
| 5 |
+
"train_samples": 5678,
|
| 6 |
+
"train_samples_per_second": 0.812,
|
| 7 |
+
"train_steps_per_second": 0.051
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2146 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 1065,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"grad_norm": 2.389887278304759,
|
| 14 |
+
"learning_rate": 4.6728971962616815e-09,
|
| 15 |
+
"logits/chosen": -2.4213736057281494,
|
| 16 |
+
"logits/rejected": -2.1724228858947754,
|
| 17 |
+
"logps/chosen": -311.7572021484375,
|
| 18 |
+
"logps/rejected": -242.86618041992188,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/margins_max": 0.0,
|
| 24 |
+
"rewards/margins_min": 0.0,
|
| 25 |
+
"rewards/margins_std": 0.0,
|
| 26 |
+
"rewards/rejected": 0.0,
|
| 27 |
+
"step": 1
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.03,
|
| 31 |
+
"grad_norm": 2.342772419823143,
|
| 32 |
+
"learning_rate": 4.672897196261682e-08,
|
| 33 |
+
"logits/chosen": -2.4386391639709473,
|
| 34 |
+
"logits/rejected": -2.2656712532043457,
|
| 35 |
+
"logps/chosen": -307.246826171875,
|
| 36 |
+
"logps/rejected": -312.6949157714844,
|
| 37 |
+
"loss": 0.6929,
|
| 38 |
+
"rewards/accuracies": 0.3333333432674408,
|
| 39 |
+
"rewards/chosen": 0.0003690579324029386,
|
| 40 |
+
"rewards/margins": 0.0005327555700205266,
|
| 41 |
+
"rewards/margins_max": 0.0024954795371741056,
|
| 42 |
+
"rewards/margins_min": -0.001210648100823164,
|
| 43 |
+
"rewards/margins_std": 0.001644498435780406,
|
| 44 |
+
"rewards/rejected": -0.0001636976667214185,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.06,
|
| 49 |
+
"grad_norm": 2.708891346495448,
|
| 50 |
+
"learning_rate": 9.345794392523364e-08,
|
| 51 |
+
"logits/chosen": -2.376209259033203,
|
| 52 |
+
"logits/rejected": -2.2393646240234375,
|
| 53 |
+
"logps/chosen": -266.4408874511719,
|
| 54 |
+
"logps/rejected": -249.82864379882812,
|
| 55 |
+
"loss": 0.6932,
|
| 56 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 57 |
+
"rewards/chosen": -4.360840466688387e-05,
|
| 58 |
+
"rewards/margins": -0.0001249962078873068,
|
| 59 |
+
"rewards/margins_max": 0.0032942264806479216,
|
| 60 |
+
"rewards/margins_min": -0.0032514198683202267,
|
| 61 |
+
"rewards/margins_std": 0.0029180804267525673,
|
| 62 |
+
"rewards/rejected": 8.138775592669845e-05,
|
| 63 |
+
"step": 20
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.08,
|
| 67 |
+
"grad_norm": 3.3723295030980913,
|
| 68 |
+
"learning_rate": 1.4018691588785045e-07,
|
| 69 |
+
"logits/chosen": -2.4367473125457764,
|
| 70 |
+
"logits/rejected": -2.2316088676452637,
|
| 71 |
+
"logps/chosen": -277.2991638183594,
|
| 72 |
+
"logps/rejected": -269.08251953125,
|
| 73 |
+
"loss": 0.6927,
|
| 74 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 75 |
+
"rewards/chosen": 0.0005795572651550174,
|
| 76 |
+
"rewards/margins": 0.001557953073643148,
|
| 77 |
+
"rewards/margins_max": 0.005679330788552761,
|
| 78 |
+
"rewards/margins_min": -0.0020338702015578747,
|
| 79 |
+
"rewards/margins_std": 0.003446707036346197,
|
| 80 |
+
"rewards/rejected": -0.0009783958084881306,
|
| 81 |
+
"step": 30
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.11,
|
| 85 |
+
"grad_norm": 2.483665988923904,
|
| 86 |
+
"learning_rate": 1.8691588785046729e-07,
|
| 87 |
+
"logits/chosen": -2.5130770206451416,
|
| 88 |
+
"logits/rejected": -2.2882275581359863,
|
| 89 |
+
"logps/chosen": -264.2230529785156,
|
| 90 |
+
"logps/rejected": -255.9278106689453,
|
| 91 |
+
"loss": 0.6922,
|
| 92 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 93 |
+
"rewards/chosen": 6.876837142044678e-05,
|
| 94 |
+
"rewards/margins": 0.0020052469335496426,
|
| 95 |
+
"rewards/margins_max": 0.006453676614910364,
|
| 96 |
+
"rewards/margins_min": -0.0016247869934886694,
|
| 97 |
+
"rewards/margins_std": 0.003689850913360715,
|
| 98 |
+
"rewards/rejected": -0.0019364787731319666,
|
| 99 |
+
"step": 40
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.14,
|
| 103 |
+
"grad_norm": 3.411713863829653,
|
| 104 |
+
"learning_rate": 2.336448598130841e-07,
|
| 105 |
+
"logits/chosen": -2.450375556945801,
|
| 106 |
+
"logits/rejected": -2.2387261390686035,
|
| 107 |
+
"logps/chosen": -256.00738525390625,
|
| 108 |
+
"logps/rejected": -272.00823974609375,
|
| 109 |
+
"loss": 0.691,
|
| 110 |
+
"rewards/accuracies": 0.762499988079071,
|
| 111 |
+
"rewards/chosen": 0.0013086812105029821,
|
| 112 |
+
"rewards/margins": 0.004717600531876087,
|
| 113 |
+
"rewards/margins_max": 0.0112897465005517,
|
| 114 |
+
"rewards/margins_min": -0.0005666814395226538,
|
| 115 |
+
"rewards/margins_std": 0.005381508264690638,
|
| 116 |
+
"rewards/rejected": -0.0034089195542037487,
|
| 117 |
+
"step": 50
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.17,
|
| 121 |
+
"grad_norm": 3.239077774946647,
|
| 122 |
+
"learning_rate": 2.803738317757009e-07,
|
| 123 |
+
"logits/chosen": -2.4118704795837402,
|
| 124 |
+
"logits/rejected": -2.242102861404419,
|
| 125 |
+
"logps/chosen": -288.9608459472656,
|
| 126 |
+
"logps/rejected": -306.24578857421875,
|
| 127 |
+
"loss": 0.6897,
|
| 128 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 129 |
+
"rewards/chosen": 0.0009396459208801389,
|
| 130 |
+
"rewards/margins": 0.007233618758618832,
|
| 131 |
+
"rewards/margins_max": 0.01480179838836193,
|
| 132 |
+
"rewards/margins_min": 0.00047727013588882983,
|
| 133 |
+
"rewards/margins_std": 0.006457436829805374,
|
| 134 |
+
"rewards/rejected": -0.006293973419815302,
|
| 135 |
+
"step": 60
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.2,
|
| 139 |
+
"grad_norm": 2.4710687298974907,
|
| 140 |
+
"learning_rate": 3.271028037383177e-07,
|
| 141 |
+
"logits/chosen": -2.4623990058898926,
|
| 142 |
+
"logits/rejected": -2.2512192726135254,
|
| 143 |
+
"logps/chosen": -285.9200134277344,
|
| 144 |
+
"logps/rejected": -274.8953552246094,
|
| 145 |
+
"loss": 0.6884,
|
| 146 |
+
"rewards/accuracies": 0.875,
|
| 147 |
+
"rewards/chosen": 0.0010064051020890474,
|
| 148 |
+
"rewards/margins": 0.008187984116375446,
|
| 149 |
+
"rewards/margins_max": 0.018993791192770004,
|
| 150 |
+
"rewards/margins_min": -0.0002981324214488268,
|
| 151 |
+
"rewards/margins_std": 0.008526557125151157,
|
| 152 |
+
"rewards/rejected": -0.007181578781455755,
|
| 153 |
+
"step": 70
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.23,
|
| 157 |
+
"grad_norm": 2.72632149666877,
|
| 158 |
+
"learning_rate": 3.7383177570093457e-07,
|
| 159 |
+
"logits/chosen": -2.520550012588501,
|
| 160 |
+
"logits/rejected": -2.3816895484924316,
|
| 161 |
+
"logps/chosen": -301.37335205078125,
|
| 162 |
+
"logps/rejected": -331.9542236328125,
|
| 163 |
+
"loss": 0.6859,
|
| 164 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 165 |
+
"rewards/chosen": 0.002128913998603821,
|
| 166 |
+
"rewards/margins": 0.015372690744698048,
|
| 167 |
+
"rewards/margins_max": 0.03150422126054764,
|
| 168 |
+
"rewards/margins_min": 0.002726194215938449,
|
| 169 |
+
"rewards/margins_std": 0.013175850734114647,
|
| 170 |
+
"rewards/rejected": -0.013243774883449078,
|
| 171 |
+
"step": 80
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.25,
|
| 175 |
+
"grad_norm": 3.532149748827834,
|
| 176 |
+
"learning_rate": 4.205607476635514e-07,
|
| 177 |
+
"logits/chosen": -2.3819632530212402,
|
| 178 |
+
"logits/rejected": -2.2091143131256104,
|
| 179 |
+
"logps/chosen": -247.6768035888672,
|
| 180 |
+
"logps/rejected": -231.6964111328125,
|
| 181 |
+
"loss": 0.6825,
|
| 182 |
+
"rewards/accuracies": 0.949999988079071,
|
| 183 |
+
"rewards/chosen": 0.003576954361051321,
|
| 184 |
+
"rewards/margins": 0.019739918410778046,
|
| 185 |
+
"rewards/margins_max": 0.043638236820697784,
|
| 186 |
+
"rewards/margins_min": 0.004001608118414879,
|
| 187 |
+
"rewards/margins_std": 0.017934972420334816,
|
| 188 |
+
"rewards/rejected": -0.016162965446710587,
|
| 189 |
+
"step": 90
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.28,
|
| 193 |
+
"grad_norm": 3.2302225335055166,
|
| 194 |
+
"learning_rate": 4.672897196261682e-07,
|
| 195 |
+
"logits/chosen": -2.382767915725708,
|
| 196 |
+
"logits/rejected": -2.2391555309295654,
|
| 197 |
+
"logps/chosen": -264.28216552734375,
|
| 198 |
+
"logps/rejected": -304.63079833984375,
|
| 199 |
+
"loss": 0.6787,
|
| 200 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 201 |
+
"rewards/chosen": 0.0004217555688228458,
|
| 202 |
+
"rewards/margins": 0.025600453838706017,
|
| 203 |
+
"rewards/margins_max": 0.054918091744184494,
|
| 204 |
+
"rewards/margins_min": 0.0039414153434336185,
|
| 205 |
+
"rewards/margins_std": 0.02377716824412346,
|
| 206 |
+
"rewards/rejected": -0.0251787006855011,
|
| 207 |
+
"step": 100
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.28,
|
| 211 |
+
"eval_logits/chosen": -2.3056893348693848,
|
| 212 |
+
"eval_logits/rejected": -2.1834657192230225,
|
| 213 |
+
"eval_logps/chosen": -276.8445739746094,
|
| 214 |
+
"eval_logps/rejected": -260.2681884765625,
|
| 215 |
+
"eval_loss": 0.6901603937149048,
|
| 216 |
+
"eval_rewards/accuracies": 0.6050000190734863,
|
| 217 |
+
"eval_rewards/chosen": -0.019638627767562866,
|
| 218 |
+
"eval_rewards/margins": 0.007838762365281582,
|
| 219 |
+
"eval_rewards/margins_max": 0.06576983630657196,
|
| 220 |
+
"eval_rewards/margins_min": -0.04385747015476227,
|
| 221 |
+
"eval_rewards/margins_std": 0.03519851341843605,
|
| 222 |
+
"eval_rewards/rejected": -0.027477389201521873,
|
| 223 |
+
"eval_runtime": 739.5177,
|
| 224 |
+
"eval_samples_per_second": 2.704,
|
| 225 |
+
"eval_steps_per_second": 0.169,
|
| 226 |
+
"step": 100
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.31,
|
| 230 |
+
"grad_norm": 3.365787507138989,
|
| 231 |
+
"learning_rate": 4.999879018839287e-07,
|
| 232 |
+
"logits/chosen": -2.63311505317688,
|
| 233 |
+
"logits/rejected": -2.3202548027038574,
|
| 234 |
+
"logps/chosen": -347.92962646484375,
|
| 235 |
+
"logps/rejected": -333.94036865234375,
|
| 236 |
+
"loss": 0.671,
|
| 237 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 238 |
+
"rewards/chosen": 0.005012608133256435,
|
| 239 |
+
"rewards/margins": 0.04701961204409599,
|
| 240 |
+
"rewards/margins_max": 0.09378845244646072,
|
| 241 |
+
"rewards/margins_min": 0.00973365269601345,
|
| 242 |
+
"rewards/margins_std": 0.03826902061700821,
|
| 243 |
+
"rewards/rejected": -0.04200700670480728,
|
| 244 |
+
"step": 110
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.34,
|
| 248 |
+
"grad_norm": 2.879990268796799,
|
| 249 |
+
"learning_rate": 4.997728568369408e-07,
|
| 250 |
+
"logits/chosen": -2.5022997856140137,
|
| 251 |
+
"logits/rejected": -2.3478662967681885,
|
| 252 |
+
"logps/chosen": -323.26910400390625,
|
| 253 |
+
"logps/rejected": -313.161865234375,
|
| 254 |
+
"loss": 0.6677,
|
| 255 |
+
"rewards/accuracies": 0.925000011920929,
|
| 256 |
+
"rewards/chosen": 0.006730080582201481,
|
| 257 |
+
"rewards/margins": 0.05950545519590378,
|
| 258 |
+
"rewards/margins_max": 0.12900672852993011,
|
| 259 |
+
"rewards/margins_min": 0.014946443028748035,
|
| 260 |
+
"rewards/margins_std": 0.05207683518528938,
|
| 261 |
+
"rewards/rejected": -0.05277537554502487,
|
| 262 |
+
"step": 120
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.37,
|
| 266 |
+
"grad_norm": 2.7102746510422207,
|
| 267 |
+
"learning_rate": 4.992892309373227e-07,
|
| 268 |
+
"logits/chosen": -2.4983277320861816,
|
| 269 |
+
"logits/rejected": -2.272451877593994,
|
| 270 |
+
"logps/chosen": -352.60589599609375,
|
| 271 |
+
"logps/rejected": -349.2779541015625,
|
| 272 |
+
"loss": 0.6585,
|
| 273 |
+
"rewards/accuracies": 1.0,
|
| 274 |
+
"rewards/chosen": 0.008741674944758415,
|
| 275 |
+
"rewards/margins": 0.07447224110364914,
|
| 276 |
+
"rewards/margins_max": 0.14192989468574524,
|
| 277 |
+
"rewards/margins_min": 0.014298012480139732,
|
| 278 |
+
"rewards/margins_std": 0.057725705206394196,
|
| 279 |
+
"rewards/rejected": -0.06573057919740677,
|
| 280 |
+
"step": 130
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.39,
|
| 284 |
+
"grad_norm": 2.2127681203334464,
|
| 285 |
+
"learning_rate": 4.985375442281968e-07,
|
| 286 |
+
"logits/chosen": -2.4224705696105957,
|
| 287 |
+
"logits/rejected": -2.2639002799987793,
|
| 288 |
+
"logps/chosen": -293.6229248046875,
|
| 289 |
+
"logps/rejected": -315.5215759277344,
|
| 290 |
+
"loss": 0.6574,
|
| 291 |
+
"rewards/accuracies": 0.925000011920929,
|
| 292 |
+
"rewards/chosen": 0.003200127277523279,
|
| 293 |
+
"rewards/margins": 0.08671971410512924,
|
| 294 |
+
"rewards/margins_max": 0.17399398982524872,
|
| 295 |
+
"rewards/margins_min": 0.014883764088153839,
|
| 296 |
+
"rewards/margins_std": 0.0733809694647789,
|
| 297 |
+
"rewards/rejected": -0.0835195854306221,
|
| 298 |
+
"step": 140
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.42,
|
| 302 |
+
"grad_norm": 2.9890965365572253,
|
| 303 |
+
"learning_rate": 4.975186049985817e-07,
|
| 304 |
+
"logits/chosen": -2.349522829055786,
|
| 305 |
+
"logits/rejected": -2.1850757598876953,
|
| 306 |
+
"logps/chosen": -268.02728271484375,
|
| 307 |
+
"logps/rejected": -252.3187713623047,
|
| 308 |
+
"loss": 0.6457,
|
| 309 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 310 |
+
"rewards/chosen": 0.00209013931453228,
|
| 311 |
+
"rewards/margins": 0.09418609738349915,
|
| 312 |
+
"rewards/margins_max": 0.189139723777771,
|
| 313 |
+
"rewards/margins_min": 0.019670698791742325,
|
| 314 |
+
"rewards/margins_std": 0.07843033969402313,
|
| 315 |
+
"rewards/rejected": -0.09209596365690231,
|
| 316 |
+
"step": 150
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.45,
|
| 320 |
+
"grad_norm": 2.5940129709794473,
|
| 321 |
+
"learning_rate": 4.962335089142375e-07,
|
| 322 |
+
"logits/chosen": -2.3253729343414307,
|
| 323 |
+
"logits/rejected": -2.1711182594299316,
|
| 324 |
+
"logps/chosen": -270.85052490234375,
|
| 325 |
+
"logps/rejected": -279.93963623046875,
|
| 326 |
+
"loss": 0.639,
|
| 327 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 328 |
+
"rewards/chosen": -0.0029645890463143587,
|
| 329 |
+
"rewards/margins": 0.11249279975891113,
|
| 330 |
+
"rewards/margins_max": 0.23242349922657013,
|
| 331 |
+
"rewards/margins_min": 0.026638973504304886,
|
| 332 |
+
"rewards/margins_std": 0.09201127290725708,
|
| 333 |
+
"rewards/rejected": -0.11545737832784653,
|
| 334 |
+
"step": 160
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.48,
|
| 338 |
+
"grad_norm": 2.9651475431583085,
|
| 339 |
+
"learning_rate": 4.946836378394966e-07,
|
| 340 |
+
"logits/chosen": -2.4186384677886963,
|
| 341 |
+
"logits/rejected": -2.2767791748046875,
|
| 342 |
+
"logps/chosen": -299.94921875,
|
| 343 |
+
"logps/rejected": -348.6862487792969,
|
| 344 |
+
"loss": 0.63,
|
| 345 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 346 |
+
"rewards/chosen": -0.002911825431510806,
|
| 347 |
+
"rewards/margins": 0.1475684940814972,
|
| 348 |
+
"rewards/margins_max": 0.2743604779243469,
|
| 349 |
+
"rewards/margins_min": 0.03711001202464104,
|
| 350 |
+
"rewards/margins_std": 0.10615728050470352,
|
| 351 |
+
"rewards/rejected": -0.15048031508922577,
|
| 352 |
+
"step": 170
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.51,
|
| 356 |
+
"grad_norm": 2.8561721958106068,
|
| 357 |
+
"learning_rate": 4.92870658351344e-07,
|
| 358 |
+
"logits/chosen": -2.456223964691162,
|
| 359 |
+
"logits/rejected": -2.226595401763916,
|
| 360 |
+
"logps/chosen": -309.79547119140625,
|
| 361 |
+
"logps/rejected": -350.2745666503906,
|
| 362 |
+
"loss": 0.6219,
|
| 363 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 364 |
+
"rewards/chosen": -0.02767277881503105,
|
| 365 |
+
"rewards/margins": 0.15511782467365265,
|
| 366 |
+
"rewards/margins_max": 0.3061321973800659,
|
| 367 |
+
"rewards/margins_min": 0.04783701151609421,
|
| 368 |
+
"rewards/margins_std": 0.11824627220630646,
|
| 369 |
+
"rewards/rejected": -0.182790607213974,
|
| 370 |
+
"step": 180
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.54,
|
| 374 |
+
"grad_norm": 3.036470774150159,
|
| 375 |
+
"learning_rate": 4.90796519947347e-07,
|
| 376 |
+
"logits/chosen": -2.3368537425994873,
|
| 377 |
+
"logits/rejected": -2.083339214324951,
|
| 378 |
+
"logps/chosen": -293.1517639160156,
|
| 379 |
+
"logps/rejected": -294.11480712890625,
|
| 380 |
+
"loss": 0.6092,
|
| 381 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 382 |
+
"rewards/chosen": -0.01847015507519245,
|
| 383 |
+
"rewards/margins": 0.18039044737815857,
|
| 384 |
+
"rewards/margins_max": 0.36516961455345154,
|
| 385 |
+
"rewards/margins_min": 0.041585661470890045,
|
| 386 |
+
"rewards/margins_std": 0.14767181873321533,
|
| 387 |
+
"rewards/rejected": -0.19886058568954468,
|
| 388 |
+
"step": 190
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.56,
|
| 392 |
+
"grad_norm": 3.196285034512815,
|
| 393 |
+
"learning_rate": 4.88463452949359e-07,
|
| 394 |
+
"logits/chosen": -2.472463607788086,
|
| 395 |
+
"logits/rejected": -2.262190103530884,
|
| 396 |
+
"logps/chosen": -323.2473449707031,
|
| 397 |
+
"logps/rejected": -331.390625,
|
| 398 |
+
"loss": 0.6038,
|
| 399 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 400 |
+
"rewards/chosen": -0.0690370723605156,
|
| 401 |
+
"rewards/margins": 0.21737924218177795,
|
| 402 |
+
"rewards/margins_max": 0.4769272804260254,
|
| 403 |
+
"rewards/margins_min": 0.030699292197823524,
|
| 404 |
+
"rewards/margins_std": 0.19959601759910583,
|
| 405 |
+
"rewards/rejected": -0.28641632199287415,
|
| 406 |
+
"step": 200
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.56,
|
| 410 |
+
"eval_logits/chosen": -2.273627519607544,
|
| 411 |
+
"eval_logits/rejected": -2.1563339233398438,
|
| 412 |
+
"eval_logps/chosen": -296.0924072265625,
|
| 413 |
+
"eval_logps/rejected": -283.1363525390625,
|
| 414 |
+
"eval_loss": 0.6828743815422058,
|
| 415 |
+
"eval_rewards/accuracies": 0.5929999947547913,
|
| 416 |
+
"eval_rewards/chosen": -0.21211715042591095,
|
| 417 |
+
"eval_rewards/margins": 0.04404139891266823,
|
| 418 |
+
"eval_rewards/margins_max": 0.41857707500457764,
|
| 419 |
+
"eval_rewards/margins_min": -0.28833016753196716,
|
| 420 |
+
"eval_rewards/margins_std": 0.2264755815267563,
|
| 421 |
+
"eval_rewards/rejected": -0.2561585307121277,
|
| 422 |
+
"eval_runtime": 740.6676,
|
| 423 |
+
"eval_samples_per_second": 2.7,
|
| 424 |
+
"eval_steps_per_second": 0.169,
|
| 425 |
+
"step": 200
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 0.59,
|
| 429 |
+
"grad_norm": 3.02540953835276,
|
| 430 |
+
"learning_rate": 4.858739661052539e-07,
|
| 431 |
+
"logits/chosen": -2.3480424880981445,
|
| 432 |
+
"logits/rejected": -2.227659225463867,
|
| 433 |
+
"logps/chosen": -289.94549560546875,
|
| 434 |
+
"logps/rejected": -328.08551025390625,
|
| 435 |
+
"loss": 0.5879,
|
| 436 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 437 |
+
"rewards/chosen": -0.05972708389163017,
|
| 438 |
+
"rewards/margins": 0.23827257752418518,
|
| 439 |
+
"rewards/margins_max": 0.48445525765419006,
|
| 440 |
+
"rewards/margins_min": 0.0391400083899498,
|
| 441 |
+
"rewards/margins_std": 0.2039383351802826,
|
| 442 |
+
"rewards/rejected": -0.29799962043762207,
|
| 443 |
+
"step": 210
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.62,
|
| 447 |
+
"grad_norm": 2.7276849485746,
|
| 448 |
+
"learning_rate": 4.830308438912687e-07,
|
| 449 |
+
"logits/chosen": -2.473395824432373,
|
| 450 |
+
"logits/rejected": -2.196845531463623,
|
| 451 |
+
"logps/chosen": -333.0096435546875,
|
| 452 |
+
"logps/rejected": -351.48321533203125,
|
| 453 |
+
"loss": 0.5788,
|
| 454 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 455 |
+
"rewards/chosen": -0.09599296748638153,
|
| 456 |
+
"rewards/margins": 0.2686784863471985,
|
| 457 |
+
"rewards/margins_max": 0.5082738995552063,
|
| 458 |
+
"rewards/margins_min": 0.044809646904468536,
|
| 459 |
+
"rewards/margins_std": 0.2064463347196579,
|
| 460 |
+
"rewards/rejected": -0.3646714687347412,
|
| 461 |
+
"step": 220
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.65,
|
| 465 |
+
"grad_norm": 2.8196760394643383,
|
| 466 |
+
"learning_rate": 4.799371435178545e-07,
|
| 467 |
+
"logits/chosen": -2.5183844566345215,
|
| 468 |
+
"logits/rejected": -2.2554872035980225,
|
| 469 |
+
"logps/chosen": -356.23358154296875,
|
| 470 |
+
"logps/rejected": -340.7899169921875,
|
| 471 |
+
"loss": 0.5631,
|
| 472 |
+
"rewards/accuracies": 0.949999988079071,
|
| 473 |
+
"rewards/chosen": -0.11530466377735138,
|
| 474 |
+
"rewards/margins": 0.29081040620803833,
|
| 475 |
+
"rewards/margins_max": 0.5872799754142761,
|
| 476 |
+
"rewards/margins_min": 0.0643739253282547,
|
| 477 |
+
"rewards/margins_std": 0.23660854995250702,
|
| 478 |
+
"rewards/rejected": -0.4061151146888733,
|
| 479 |
+
"step": 230
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.68,
|
| 483 |
+
"grad_norm": 2.402815708848366,
|
| 484 |
+
"learning_rate": 4.765961916422574e-07,
|
| 485 |
+
"logits/chosen": -2.409052610397339,
|
| 486 |
+
"logits/rejected": -2.2241270542144775,
|
| 487 |
+
"logps/chosen": -325.99786376953125,
|
| 488 |
+
"logps/rejected": -373.8414611816406,
|
| 489 |
+
"loss": 0.5776,
|
| 490 |
+
"rewards/accuracies": 0.949999988079071,
|
| 491 |
+
"rewards/chosen": -0.1929856687784195,
|
| 492 |
+
"rewards/margins": 0.2687362730503082,
|
| 493 |
+
"rewards/margins_max": 0.5277136564254761,
|
| 494 |
+
"rewards/margins_min": 0.040412187576293945,
|
| 495 |
+
"rewards/margins_std": 0.22582125663757324,
|
| 496 |
+
"rewards/rejected": -0.46172189712524414,
|
| 497 |
+
"step": 240
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 0.7,
|
| 501 |
+
"grad_norm": 2.9799913096461217,
|
| 502 |
+
"learning_rate": 4.730115807913626e-07,
|
| 503 |
+
"logits/chosen": -2.446721076965332,
|
| 504 |
+
"logits/rejected": -2.2028250694274902,
|
| 505 |
+
"logps/chosen": -303.15399169921875,
|
| 506 |
+
"logps/rejected": -335.6115417480469,
|
| 507 |
+
"loss": 0.5469,
|
| 508 |
+
"rewards/accuracies": 0.925000011920929,
|
| 509 |
+
"rewards/chosen": -0.16406993567943573,
|
| 510 |
+
"rewards/margins": 0.3364391326904297,
|
| 511 |
+
"rewards/margins_max": 0.7914501428604126,
|
| 512 |
+
"rewards/margins_min": 0.05388214439153671,
|
| 513 |
+
"rewards/margins_std": 0.33590734004974365,
|
| 514 |
+
"rewards/rejected": -0.5005090832710266,
|
| 515 |
+
"step": 250
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.73,
|
| 519 |
+
"grad_norm": 2.770663569324122,
|
| 520 |
+
"learning_rate": 4.691871654986485e-07,
|
| 521 |
+
"logits/chosen": -2.3937630653381348,
|
| 522 |
+
"logits/rejected": -2.21606183052063,
|
| 523 |
+
"logps/chosen": -288.9862060546875,
|
| 524 |
+
"logps/rejected": -352.5678405761719,
|
| 525 |
+
"loss": 0.536,
|
| 526 |
+
"rewards/accuracies": 0.925000011920929,
|
| 527 |
+
"rewards/chosen": -0.2037801295518875,
|
| 528 |
+
"rewards/margins": 0.34631386399269104,
|
| 529 |
+
"rewards/margins_max": 0.792448103427887,
|
| 530 |
+
"rewards/margins_min": 0.02828853949904442,
|
| 531 |
+
"rewards/margins_std": 0.3575611412525177,
|
| 532 |
+
"rewards/rejected": -0.5500940084457397,
|
| 533 |
+
"step": 260
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.76,
|
| 537 |
+
"grad_norm": 3.098378120195792,
|
| 538 |
+
"learning_rate": 4.6512705815940536e-07,
|
| 539 |
+
"logits/chosen": -2.3696112632751465,
|
| 540 |
+
"logits/rejected": -2.18962025642395,
|
| 541 |
+
"logps/chosen": -306.8760070800781,
|
| 542 |
+
"logps/rejected": -372.0642395019531,
|
| 543 |
+
"loss": 0.5394,
|
| 544 |
+
"rewards/accuracies": 0.887499988079071,
|
| 545 |
+
"rewards/chosen": -0.2417019158601761,
|
| 546 |
+
"rewards/margins": 0.41768041253089905,
|
| 547 |
+
"rewards/margins_max": 0.8514153361320496,
|
| 548 |
+
"rewards/margins_min": 0.05087714642286301,
|
| 549 |
+
"rewards/margins_std": 0.3643653392791748,
|
| 550 |
+
"rewards/rejected": -0.6593823432922363,
|
| 551 |
+
"step": 270
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 0.79,
|
| 555 |
+
"grad_norm": 2.966546455548068,
|
| 556 |
+
"learning_rate": 4.6083562460867544e-07,
|
| 557 |
+
"logits/chosen": -2.3208565711975098,
|
| 558 |
+
"logits/rejected": -2.2000744342803955,
|
| 559 |
+
"logps/chosen": -313.9183349609375,
|
| 560 |
+
"logps/rejected": -385.6065368652344,
|
| 561 |
+
"loss": 0.5172,
|
| 562 |
+
"rewards/accuracies": 0.925000011920929,
|
| 563 |
+
"rewards/chosen": -0.2830614447593689,
|
| 564 |
+
"rewards/margins": 0.4400337338447571,
|
| 565 |
+
"rewards/margins_max": 0.9521900415420532,
|
| 566 |
+
"rewards/margins_min": 0.09125441312789917,
|
| 567 |
+
"rewards/margins_std": 0.40178269147872925,
|
| 568 |
+
"rewards/rejected": -0.7230951189994812,
|
| 569 |
+
"step": 280
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.82,
|
| 573 |
+
"grad_norm": 3.1108063896750258,
|
| 574 |
+
"learning_rate": 4.563174794266683e-07,
|
| 575 |
+
"logits/chosen": -2.305835008621216,
|
| 576 |
+
"logits/rejected": -2.194869041442871,
|
| 577 |
+
"logps/chosen": -294.03326416015625,
|
| 578 |
+
"logps/rejected": -368.85650634765625,
|
| 579 |
+
"loss": 0.5334,
|
| 580 |
+
"rewards/accuracies": 0.875,
|
| 581 |
+
"rewards/chosen": -0.28016966581344604,
|
| 582 |
+
"rewards/margins": 0.4169536232948303,
|
| 583 |
+
"rewards/margins_max": 0.9842053651809692,
|
| 584 |
+
"rewards/margins_min": 0.05335196107625961,
|
| 585 |
+
"rewards/margins_std": 0.42577672004699707,
|
| 586 |
+
"rewards/rejected": -0.6971233487129211,
|
| 587 |
+
"step": 290
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.85,
|
| 591 |
+
"grad_norm": 3.4198160465582594,
|
| 592 |
+
"learning_rate": 4.515774809767012e-07,
|
| 593 |
+
"logits/chosen": -2.4141030311584473,
|
| 594 |
+
"logits/rejected": -2.219494581222534,
|
| 595 |
+
"logps/chosen": -346.39154052734375,
|
| 596 |
+
"logps/rejected": -413.7733459472656,
|
| 597 |
+
"loss": 0.4746,
|
| 598 |
+
"rewards/accuracies": 0.949999988079071,
|
| 599 |
+
"rewards/chosen": -0.3215632438659668,
|
| 600 |
+
"rewards/margins": 0.559494137763977,
|
| 601 |
+
"rewards/margins_max": 1.2075681686401367,
|
| 602 |
+
"rewards/margins_min": 0.1119498759508133,
|
| 603 |
+
"rewards/margins_std": 0.4947145879268646,
|
| 604 |
+
"rewards/rejected": -0.8810573816299438,
|
| 605 |
+
"step": 300
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.85,
|
| 609 |
+
"eval_logits/chosen": -2.1990644931793213,
|
| 610 |
+
"eval_logits/rejected": -2.090372085571289,
|
| 611 |
+
"eval_logps/chosen": -352.614013671875,
|
| 612 |
+
"eval_logps/rejected": -342.9794921875,
|
| 613 |
+
"eval_loss": 0.7105380892753601,
|
| 614 |
+
"eval_rewards/accuracies": 0.5659999847412109,
|
| 615 |
+
"eval_rewards/chosen": -0.7773330211639404,
|
| 616 |
+
"eval_rewards/margins": 0.07725735753774643,
|
| 617 |
+
"eval_rewards/margins_max": 1.040091872215271,
|
| 618 |
+
"eval_rewards/margins_min": -0.8433572053909302,
|
| 619 |
+
"eval_rewards/margins_std": 0.6093021035194397,
|
| 620 |
+
"eval_rewards/rejected": -0.8545902967453003,
|
| 621 |
+
"eval_runtime": 741.0262,
|
| 622 |
+
"eval_samples_per_second": 2.699,
|
| 623 |
+
"eval_steps_per_second": 0.169,
|
| 624 |
+
"step": 300
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.87,
|
| 628 |
+
"grad_norm": 2.5169015808444857,
|
| 629 |
+
"learning_rate": 4.4662072618099887e-07,
|
| 630 |
+
"logits/chosen": -2.445680856704712,
|
| 631 |
+
"logits/rejected": -2.1884543895721436,
|
| 632 |
+
"logps/chosen": -313.87322998046875,
|
| 633 |
+
"logps/rejected": -332.14422607421875,
|
| 634 |
+
"loss": 0.5121,
|
| 635 |
+
"rewards/accuracies": 0.887499988079071,
|
| 636 |
+
"rewards/chosen": -0.4068140387535095,
|
| 637 |
+
"rewards/margins": 0.4162236154079437,
|
| 638 |
+
"rewards/margins_max": 0.9356614947319031,
|
| 639 |
+
"rewards/margins_min": 0.056422315537929535,
|
| 640 |
+
"rewards/margins_std": 0.3940203785896301,
|
| 641 |
+
"rewards/rejected": -0.8230376243591309,
|
| 642 |
+
"step": 310
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.9,
|
| 646 |
+
"grad_norm": 3.3837390362012707,
|
| 647 |
+
"learning_rate": 4.414525450399712e-07,
|
| 648 |
+
"logits/chosen": -2.325852632522583,
|
| 649 |
+
"logits/rejected": -2.133544683456421,
|
| 650 |
+
"logps/chosen": -339.6539611816406,
|
| 651 |
+
"logps/rejected": -396.7124328613281,
|
| 652 |
+
"loss": 0.4964,
|
| 653 |
+
"rewards/accuracies": 0.875,
|
| 654 |
+
"rewards/chosen": -0.48702773451805115,
|
| 655 |
+
"rewards/margins": 0.5068559050559998,
|
| 656 |
+
"rewards/margins_max": 1.1567133665084839,
|
| 657 |
+
"rewards/margins_min": -0.029241245239973068,
|
| 658 |
+
"rewards/margins_std": 0.5536248683929443,
|
| 659 |
+
"rewards/rejected": -0.9938837289810181,
|
| 660 |
+
"step": 320
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.93,
|
| 664 |
+
"grad_norm": 3.325411841334806,
|
| 665 |
+
"learning_rate": 4.360784949008615e-07,
|
| 666 |
+
"logits/chosen": -2.4802184104919434,
|
| 667 |
+
"logits/rejected": -2.291903018951416,
|
| 668 |
+
"logps/chosen": -379.33331298828125,
|
| 669 |
+
"logps/rejected": -424.1192932128906,
|
| 670 |
+
"loss": 0.4802,
|
| 671 |
+
"rewards/accuracies": 0.987500011920929,
|
| 672 |
+
"rewards/chosen": -0.49472981691360474,
|
| 673 |
+
"rewards/margins": 0.692683756351471,
|
| 674 |
+
"rewards/margins_max": 1.5269715785980225,
|
| 675 |
+
"rewards/margins_min": 0.13377800583839417,
|
| 676 |
+
"rewards/margins_std": 0.6406728029251099,
|
| 677 |
+
"rewards/rejected": -1.1874135732650757,
|
| 678 |
+
"step": 330
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 0.96,
|
| 682 |
+
"grad_norm": 3.1486731504941217,
|
| 683 |
+
"learning_rate": 4.305043544819289e-07,
|
| 684 |
+
"logits/chosen": -2.2821261882781982,
|
| 685 |
+
"logits/rejected": -2.1466071605682373,
|
| 686 |
+
"logps/chosen": -322.30230712890625,
|
| 687 |
+
"logps/rejected": -388.6499328613281,
|
| 688 |
+
"loss": 0.4944,
|
| 689 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 690 |
+
"rewards/chosen": -0.4555594027042389,
|
| 691 |
+
"rewards/margins": 0.5605371594429016,
|
| 692 |
+
"rewards/margins_max": 1.1739370822906494,
|
| 693 |
+
"rewards/margins_min": 0.10414503514766693,
|
| 694 |
+
"rewards/margins_std": 0.4933851659297943,
|
| 695 |
+
"rewards/rejected": -1.016096591949463,
|
| 696 |
+
"step": 340
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.99,
|
| 700 |
+
"grad_norm": 3.4872610687117387,
|
| 701 |
+
"learning_rate": 4.247361176585903e-07,
|
| 702 |
+
"logits/chosen": -2.2431674003601074,
|
| 703 |
+
"logits/rejected": -2.1132471561431885,
|
| 704 |
+
"logps/chosen": -330.2970886230469,
|
| 705 |
+
"logps/rejected": -435.4002990722656,
|
| 706 |
+
"loss": 0.4735,
|
| 707 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 708 |
+
"rewards/chosen": -0.48258066177368164,
|
| 709 |
+
"rewards/margins": 0.6382626295089722,
|
| 710 |
+
"rewards/margins_max": 1.3624789714813232,
|
| 711 |
+
"rewards/margins_min": 0.10766267776489258,
|
| 712 |
+
"rewards/margins_std": 0.5704151391983032,
|
| 713 |
+
"rewards/rejected": -1.1208432912826538,
|
| 714 |
+
"step": 350
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.01,
|
| 718 |
+
"grad_norm": 3.207929181161526,
|
| 719 |
+
"learning_rate": 4.187799870182038e-07,
|
| 720 |
+
"logits/chosen": -2.299013376235962,
|
| 721 |
+
"logits/rejected": -2.159662961959839,
|
| 722 |
+
"logps/chosen": -329.9014892578125,
|
| 723 |
+
"logps/rejected": -411.421875,
|
| 724 |
+
"loss": 0.4418,
|
| 725 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 726 |
+
"rewards/chosen": -0.49954190850257874,
|
| 727 |
+
"rewards/margins": 0.6922783255577087,
|
| 728 |
+
"rewards/margins_max": 1.497651219367981,
|
| 729 |
+
"rewards/margins_min": 0.14925283193588257,
|
| 730 |
+
"rewards/margins_std": 0.6044853925704956,
|
| 731 |
+
"rewards/rejected": -1.1918203830718994,
|
| 732 |
+
"step": 360
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 1.04,
|
| 736 |
+
"grad_norm": 4.274237019230171,
|
| 737 |
+
"learning_rate": 4.126423671904236e-07,
|
| 738 |
+
"logits/chosen": -2.34145188331604,
|
| 739 |
+
"logits/rejected": -2.116023063659668,
|
| 740 |
+
"logps/chosen": -384.4696960449219,
|
| 741 |
+
"logps/rejected": -413.1422424316406,
|
| 742 |
+
"loss": 0.4731,
|
| 743 |
+
"rewards/accuracies": 0.887499988079071,
|
| 744 |
+
"rewards/chosen": -0.616447925567627,
|
| 745 |
+
"rewards/margins": 0.7246918678283691,
|
| 746 |
+
"rewards/margins_max": 1.6939563751220703,
|
| 747 |
+
"rewards/margins_min": -0.03321857377886772,
|
| 748 |
+
"rewards/margins_std": 0.7812119722366333,
|
| 749 |
+
"rewards/rejected": -1.341139793395996,
|
| 750 |
+
"step": 370
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 1.07,
|
| 754 |
+
"grad_norm": 3.3134994438588827,
|
| 755 |
+
"learning_rate": 4.0632985796030007e-07,
|
| 756 |
+
"logits/chosen": -2.20294451713562,
|
| 757 |
+
"logits/rejected": -2.036015033721924,
|
| 758 |
+
"logps/chosen": -314.45513916015625,
|
| 759 |
+
"logps/rejected": -395.52667236328125,
|
| 760 |
+
"loss": 0.4449,
|
| 761 |
+
"rewards/accuracies": 0.925000011920929,
|
| 762 |
+
"rewards/chosen": -0.549362063407898,
|
| 763 |
+
"rewards/margins": 0.6994296312332153,
|
| 764 |
+
"rewards/margins_max": 1.5058152675628662,
|
| 765 |
+
"rewards/margins_min": 0.12678512930870056,
|
| 766 |
+
"rewards/margins_std": 0.6280149817466736,
|
| 767 |
+
"rewards/rejected": -1.2487916946411133,
|
| 768 |
+
"step": 380
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 1.1,
|
| 772 |
+
"grad_norm": 2.772777984001906,
|
| 773 |
+
"learning_rate": 3.9984924717152713e-07,
|
| 774 |
+
"logits/chosen": -2.216956853866577,
|
| 775 |
+
"logits/rejected": -2.116546154022217,
|
| 776 |
+
"logps/chosen": -351.6098937988281,
|
| 777 |
+
"logps/rejected": -451.20562744140625,
|
| 778 |
+
"loss": 0.4402,
|
| 779 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 780 |
+
"rewards/chosen": -0.7173673510551453,
|
| 781 |
+
"rewards/margins": 0.8140894174575806,
|
| 782 |
+
"rewards/margins_max": 1.774478554725647,
|
| 783 |
+
"rewards/margins_min": 0.1617562621831894,
|
| 784 |
+
"rewards/margins_std": 0.7393523454666138,
|
| 785 |
+
"rewards/rejected": -1.531456708908081,
|
| 786 |
+
"step": 390
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 1.13,
|
| 790 |
+
"grad_norm": 3.7243228476083776,
|
| 791 |
+
"learning_rate": 3.932075034274723e-07,
|
| 792 |
+
"logits/chosen": -2.2378880977630615,
|
| 793 |
+
"logits/rejected": -2.064462184906006,
|
| 794 |
+
"logps/chosen": -347.496337890625,
|
| 795 |
+
"logps/rejected": -426.29315185546875,
|
| 796 |
+
"loss": 0.4288,
|
| 797 |
+
"rewards/accuracies": 0.875,
|
| 798 |
+
"rewards/chosen": -0.7525204420089722,
|
| 799 |
+
"rewards/margins": 0.7026923894882202,
|
| 800 |
+
"rewards/margins_max": 1.5236282348632812,
|
| 801 |
+
"rewards/margins_min": 0.09952196478843689,
|
| 802 |
+
"rewards/margins_std": 0.647875189781189,
|
| 803 |
+
"rewards/rejected": -1.4552127122879028,
|
| 804 |
+
"step": 400
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.13,
|
| 808 |
+
"eval_logits/chosen": -2.1253983974456787,
|
| 809 |
+
"eval_logits/rejected": -2.0236704349517822,
|
| 810 |
+
"eval_logps/chosen": -409.9260559082031,
|
| 811 |
+
"eval_logps/rejected": -405.01422119140625,
|
| 812 |
+
"eval_loss": 0.7566000819206238,
|
| 813 |
+
"eval_rewards/accuracies": 0.5699999928474426,
|
| 814 |
+
"eval_rewards/chosen": -1.350453495979309,
|
| 815 |
+
"eval_rewards/margins": 0.12448413670063019,
|
| 816 |
+
"eval_rewards/margins_max": 1.6613168716430664,
|
| 817 |
+
"eval_rewards/margins_min": -1.3515390157699585,
|
| 818 |
+
"eval_rewards/margins_std": 0.9884259700775146,
|
| 819 |
+
"eval_rewards/rejected": -1.4749376773834229,
|
| 820 |
+
"eval_runtime": 740.8805,
|
| 821 |
+
"eval_samples_per_second": 2.699,
|
| 822 |
+
"eval_steps_per_second": 0.169,
|
| 823 |
+
"step": 400
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.15,
|
| 827 |
+
"grad_norm": 3.709048912929159,
|
| 828 |
+
"learning_rate": 3.8641176859783383e-07,
|
| 829 |
+
"logits/chosen": -2.263033151626587,
|
| 830 |
+
"logits/rejected": -2.110279083251953,
|
| 831 |
+
"logps/chosen": -360.94451904296875,
|
| 832 |
+
"logps/rejected": -447.3453674316406,
|
| 833 |
+
"loss": 0.446,
|
| 834 |
+
"rewards/accuracies": 0.887499988079071,
|
| 835 |
+
"rewards/chosen": -0.8266700506210327,
|
| 836 |
+
"rewards/margins": 0.7980839014053345,
|
| 837 |
+
"rewards/margins_max": 1.9258846044540405,
|
| 838 |
+
"rewards/margins_min": 0.028872152790427208,
|
| 839 |
+
"rewards/margins_std": 0.8716685175895691,
|
| 840 |
+
"rewards/rejected": -1.6247539520263672,
|
| 841 |
+
"step": 410
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 1.18,
|
| 845 |
+
"grad_norm": 3.1370524113769718,
|
| 846 |
+
"learning_rate": 3.7946935013898606e-07,
|
| 847 |
+
"logits/chosen": -2.22702693939209,
|
| 848 |
+
"logits/rejected": -2.058694839477539,
|
| 849 |
+
"logps/chosen": -378.8398742675781,
|
| 850 |
+
"logps/rejected": -486.36895751953125,
|
| 851 |
+
"loss": 0.4121,
|
| 852 |
+
"rewards/accuracies": 0.925000011920929,
|
| 853 |
+
"rewards/chosen": -0.867613673210144,
|
| 854 |
+
"rewards/margins": 0.9479442834854126,
|
| 855 |
+
"rewards/margins_max": 1.9218909740447998,
|
| 856 |
+
"rewards/margins_min": 0.206168532371521,
|
| 857 |
+
"rewards/margins_std": 0.7779111862182617,
|
| 858 |
+
"rewards/rejected": -1.815557837486267,
|
| 859 |
+
"step": 420
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.21,
|
| 863 |
+
"grad_norm": 5.732961537629863,
|
| 864 |
+
"learning_rate": 3.7238771323626817e-07,
|
| 865 |
+
"logits/chosen": -2.271839141845703,
|
| 866 |
+
"logits/rejected": -2.0797672271728516,
|
| 867 |
+
"logps/chosen": -428.86376953125,
|
| 868 |
+
"logps/rejected": -498.2832946777344,
|
| 869 |
+
"loss": 0.4013,
|
| 870 |
+
"rewards/accuracies": 0.824999988079071,
|
| 871 |
+
"rewards/chosen": -1.0997209548950195,
|
| 872 |
+
"rewards/margins": 0.86876380443573,
|
| 873 |
+
"rewards/margins_max": 1.956595778465271,
|
| 874 |
+
"rewards/margins_min": -0.09084595739841461,
|
| 875 |
+
"rewards/margins_std": 0.9087589383125305,
|
| 876 |
+
"rewards/rejected": -1.96848464012146,
|
| 877 |
+
"step": 430
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 1.24,
|
| 881 |
+
"grad_norm": 3.2379293031944587,
|
| 882 |
+
"learning_rate": 3.651744727766676e-07,
|
| 883 |
+
"logits/chosen": -2.2412235736846924,
|
| 884 |
+
"logits/rejected": -2.06864333152771,
|
| 885 |
+
"logps/chosen": -367.8034362792969,
|
| 886 |
+
"logps/rejected": -472.5411071777344,
|
| 887 |
+
"loss": 0.4283,
|
| 888 |
+
"rewards/accuracies": 0.875,
|
| 889 |
+
"rewards/chosen": -0.9834961891174316,
|
| 890 |
+
"rewards/margins": 0.9132775068283081,
|
| 891 |
+
"rewards/margins_max": 1.9850295782089233,
|
| 892 |
+
"rewards/margins_min": 0.06963314116001129,
|
| 893 |
+
"rewards/margins_std": 0.8529081344604492,
|
| 894 |
+
"rewards/rejected": -1.8967736959457397,
|
| 895 |
+
"step": 440
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 1.27,
|
| 899 |
+
"grad_norm": 3.679847665204543,
|
| 900 |
+
"learning_rate": 3.5783738516052897e-07,
|
| 901 |
+
"logits/chosen": -2.1635823249816895,
|
| 902 |
+
"logits/rejected": -2.008791208267212,
|
| 903 |
+
"logps/chosen": -370.2854919433594,
|
| 904 |
+
"logps/rejected": -473.046630859375,
|
| 905 |
+
"loss": 0.3927,
|
| 906 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 907 |
+
"rewards/chosen": -0.8544079065322876,
|
| 908 |
+
"rewards/margins": 0.9354113340377808,
|
| 909 |
+
"rewards/margins_max": 2.0640413761138916,
|
| 910 |
+
"rewards/margins_min": 0.09445329755544662,
|
| 911 |
+
"rewards/margins_std": 0.8850519061088562,
|
| 912 |
+
"rewards/rejected": -1.789819359779358,
|
| 913 |
+
"step": 450
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 1.3,
|
| 917 |
+
"grad_norm": 3.9802612609991503,
|
| 918 |
+
"learning_rate": 3.5038433996109404e-07,
|
| 919 |
+
"logits/chosen": -2.3022890090942383,
|
| 920 |
+
"logits/rejected": -2.148092269897461,
|
| 921 |
+
"logps/chosen": -420.42999267578125,
|
| 922 |
+
"logps/rejected": -540.9712524414062,
|
| 923 |
+
"loss": 0.379,
|
| 924 |
+
"rewards/accuracies": 0.925000011920929,
|
| 925 |
+
"rewards/chosen": -0.9502747654914856,
|
| 926 |
+
"rewards/margins": 1.1226153373718262,
|
| 927 |
+
"rewards/margins_max": 2.0989441871643066,
|
| 928 |
+
"rewards/margins_min": 0.2553909122943878,
|
| 929 |
+
"rewards/margins_std": 0.8486478924751282,
|
| 930 |
+
"rewards/rejected": -2.072890281677246,
|
| 931 |
+
"step": 460
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 1.32,
|
| 935 |
+
"grad_norm": 3.153480145327942,
|
| 936 |
+
"learning_rate": 3.428233514408398e-07,
|
| 937 |
+
"logits/chosen": -2.232956647872925,
|
| 938 |
+
"logits/rejected": -2.0882651805877686,
|
| 939 |
+
"logps/chosen": -383.5642395019531,
|
| 940 |
+
"logps/rejected": -496.1742248535156,
|
| 941 |
+
"loss": 0.3923,
|
| 942 |
+
"rewards/accuracies": 0.9375,
|
| 943 |
+
"rewards/chosen": -0.9364080429077148,
|
| 944 |
+
"rewards/margins": 1.1441760063171387,
|
| 945 |
+
"rewards/margins_max": 2.6022720336914062,
|
| 946 |
+
"rewards/margins_min": 0.20014441013336182,
|
| 947 |
+
"rewards/margins_std": 1.0920383930206299,
|
| 948 |
+
"rewards/rejected": -2.0805840492248535,
|
| 949 |
+
"step": 470
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 1.35,
|
| 953 |
+
"grad_norm": 3.401868329298301,
|
| 954 |
+
"learning_rate": 3.3516254993373945e-07,
|
| 955 |
+
"logits/chosen": -2.2222964763641357,
|
| 956 |
+
"logits/rejected": -2.030994176864624,
|
| 957 |
+
"logps/chosen": -403.7270812988281,
|
| 958 |
+
"logps/rejected": -489.4800720214844,
|
| 959 |
+
"loss": 0.4117,
|
| 960 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 961 |
+
"rewards/chosen": -1.032002568244934,
|
| 962 |
+
"rewards/margins": 1.0399589538574219,
|
| 963 |
+
"rewards/margins_max": 2.25704026222229,
|
| 964 |
+
"rewards/margins_min": 0.06191538646817207,
|
| 965 |
+
"rewards/margins_std": 0.9929068684577942,
|
| 966 |
+
"rewards/rejected": -2.0719614028930664,
|
| 967 |
+
"step": 480
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 1.38,
|
| 971 |
+
"grad_norm": 3.3964950447287845,
|
| 972 |
+
"learning_rate": 3.274101731027105e-07,
|
| 973 |
+
"logits/chosen": -2.2032177448272705,
|
| 974 |
+
"logits/rejected": -2.071004867553711,
|
| 975 |
+
"logps/chosen": -400.43402099609375,
|
| 976 |
+
"logps/rejected": -543.1834106445312,
|
| 977 |
+
"loss": 0.3817,
|
| 978 |
+
"rewards/accuracies": 0.875,
|
| 979 |
+
"rewards/chosen": -0.9758344888687134,
|
| 980 |
+
"rewards/margins": 1.1685841083526611,
|
| 981 |
+
"rewards/margins_max": 2.530055046081543,
|
| 982 |
+
"rewards/margins_min": 0.010178548283874989,
|
| 983 |
+
"rewards/margins_std": 1.1102135181427002,
|
| 984 |
+
"rewards/rejected": -2.144418716430664,
|
| 985 |
+
"step": 490
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 1.41,
|
| 989 |
+
"grad_norm": 3.647023353729911,
|
| 990 |
+
"learning_rate": 3.1957455708165314e-07,
|
| 991 |
+
"logits/chosen": -2.1546573638916016,
|
| 992 |
+
"logits/rejected": -2.0752437114715576,
|
| 993 |
+
"logps/chosen": -373.78045654296875,
|
| 994 |
+
"logps/rejected": -493.3614807128906,
|
| 995 |
+
"loss": 0.3807,
|
| 996 |
+
"rewards/accuracies": 0.887499988079071,
|
| 997 |
+
"rewards/chosen": -1.0127661228179932,
|
| 998 |
+
"rewards/margins": 1.0738728046417236,
|
| 999 |
+
"rewards/margins_max": 2.1911089420318604,
|
| 1000 |
+
"rewards/margins_min": 0.05973333120346069,
|
| 1001 |
+
"rewards/margins_std": 0.9444160461425781,
|
| 1002 |
+
"rewards/rejected": -2.086638927459717,
|
| 1003 |
+
"step": 500
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 1.41,
|
| 1007 |
+
"eval_logits/chosen": -2.0583739280700684,
|
| 1008 |
+
"eval_logits/rejected": -1.9637471437454224,
|
| 1009 |
+
"eval_logps/chosen": -451.7817077636719,
|
| 1010 |
+
"eval_logps/rejected": -455.107666015625,
|
| 1011 |
+
"eval_loss": 0.7769566178321838,
|
| 1012 |
+
"eval_rewards/accuracies": 0.5759999752044678,
|
| 1013 |
+
"eval_rewards/chosen": -1.769010305404663,
|
| 1014 |
+
"eval_rewards/margins": 0.20686142146587372,
|
| 1015 |
+
"eval_rewards/margins_max": 2.146589994430542,
|
| 1016 |
+
"eval_rewards/margins_min": -1.6287314891815186,
|
| 1017 |
+
"eval_rewards/margins_std": 1.2536591291427612,
|
| 1018 |
+
"eval_rewards/rejected": -1.9758716821670532,
|
| 1019 |
+
"eval_runtime": 741.3629,
|
| 1020 |
+
"eval_samples_per_second": 2.698,
|
| 1021 |
+
"eval_steps_per_second": 0.169,
|
| 1022 |
+
"step": 500
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 1.44,
|
| 1026 |
+
"grad_norm": 4.75515226711996,
|
| 1027 |
+
"learning_rate": 3.116641275116018e-07,
|
| 1028 |
+
"logits/chosen": -2.1037933826446533,
|
| 1029 |
+
"logits/rejected": -1.8988920450210571,
|
| 1030 |
+
"logps/chosen": -431.19134521484375,
|
| 1031 |
+
"logps/rejected": -524.6165161132812,
|
| 1032 |
+
"loss": 0.4047,
|
| 1033 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1034 |
+
"rewards/chosen": -1.1405599117279053,
|
| 1035 |
+
"rewards/margins": 1.1359269618988037,
|
| 1036 |
+
"rewards/margins_max": 2.8867573738098145,
|
| 1037 |
+
"rewards/margins_min": 0.03959502652287483,
|
| 1038 |
+
"rewards/margins_std": 1.2942934036254883,
|
| 1039 |
+
"rewards/rejected": -2.276486873626709,
|
| 1040 |
+
"step": 510
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 1.46,
|
| 1044 |
+
"grad_norm": 3.8133006271435814,
|
| 1045 |
+
"learning_rate": 3.036873904806295e-07,
|
| 1046 |
+
"logits/chosen": -2.175333261489868,
|
| 1047 |
+
"logits/rejected": -1.971325159072876,
|
| 1048 |
+
"logps/chosen": -386.5411682128906,
|
| 1049 |
+
"logps/rejected": -497.42767333984375,
|
| 1050 |
+
"loss": 0.3576,
|
| 1051 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1052 |
+
"rewards/chosen": -1.1177667379379272,
|
| 1053 |
+
"rewards/margins": 1.1489231586456299,
|
| 1054 |
+
"rewards/margins_max": 2.5013482570648193,
|
| 1055 |
+
"rewards/margins_min": 0.12063050270080566,
|
| 1056 |
+
"rewards/margins_std": 1.0784401893615723,
|
| 1057 |
+
"rewards/rejected": -2.2666897773742676,
|
| 1058 |
+
"step": 520
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.49,
|
| 1062 |
+
"grad_norm": 4.402083137874476,
|
| 1063 |
+
"learning_rate": 2.956529233772492e-07,
|
| 1064 |
+
"logits/chosen": -2.164149284362793,
|
| 1065 |
+
"logits/rejected": -2.0270886421203613,
|
| 1066 |
+
"logps/chosen": -445.92962646484375,
|
| 1067 |
+
"logps/rejected": -623.1121826171875,
|
| 1068 |
+
"loss": 0.3375,
|
| 1069 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1070 |
+
"rewards/chosen": -1.3388593196868896,
|
| 1071 |
+
"rewards/margins": 1.5283788442611694,
|
| 1072 |
+
"rewards/margins_max": 3.114734649658203,
|
| 1073 |
+
"rewards/margins_min": 0.2940402626991272,
|
| 1074 |
+
"rewards/margins_std": 1.2590672969818115,
|
| 1075 |
+
"rewards/rejected": -2.8672380447387695,
|
| 1076 |
+
"step": 530
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 1.52,
|
| 1080 |
+
"grad_norm": 5.278491140419052,
|
| 1081 |
+
"learning_rate": 2.875693656671431e-07,
|
| 1082 |
+
"logits/chosen": -2.153266191482544,
|
| 1083 |
+
"logits/rejected": -1.9573100805282593,
|
| 1084 |
+
"logps/chosen": -465.7792053222656,
|
| 1085 |
+
"logps/rejected": -541.7371826171875,
|
| 1086 |
+
"loss": 0.3631,
|
| 1087 |
+
"rewards/accuracies": 0.875,
|
| 1088 |
+
"rewards/chosen": -1.4917871952056885,
|
| 1089 |
+
"rewards/margins": 1.1456191539764404,
|
| 1090 |
+
"rewards/margins_max": 2.534468173980713,
|
| 1091 |
+
"rewards/margins_min": 0.050025321543216705,
|
| 1092 |
+
"rewards/margins_std": 1.114137887954712,
|
| 1093 |
+
"rewards/rejected": -2.637406349182129,
|
| 1094 |
+
"step": 540
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 1.55,
|
| 1098 |
+
"grad_norm": 2.831099826627899,
|
| 1099 |
+
"learning_rate": 2.794454096031429e-07,
|
| 1100 |
+
"logits/chosen": -2.1867547035217285,
|
| 1101 |
+
"logits/rejected": -2.0559072494506836,
|
| 1102 |
+
"logps/chosen": -509.72802734375,
|
| 1103 |
+
"logps/rejected": -703.0756225585938,
|
| 1104 |
+
"loss": 0.3108,
|
| 1105 |
+
"rewards/accuracies": 0.9375,
|
| 1106 |
+
"rewards/chosen": -1.6035360097885132,
|
| 1107 |
+
"rewards/margins": 1.853695273399353,
|
| 1108 |
+
"rewards/margins_max": 4.070513725280762,
|
| 1109 |
+
"rewards/margins_min": 0.37196239829063416,
|
| 1110 |
+
"rewards/margins_std": 1.6610209941864014,
|
| 1111 |
+
"rewards/rejected": -3.457231044769287,
|
| 1112 |
+
"step": 550
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 1.58,
|
| 1116 |
+
"grad_norm": 4.413758285977071,
|
| 1117 |
+
"learning_rate": 2.7128979087844593e-07,
|
| 1118 |
+
"logits/chosen": -2.15148663520813,
|
| 1119 |
+
"logits/rejected": -1.9961950778961182,
|
| 1120 |
+
"logps/chosen": -422.4507751464844,
|
| 1121 |
+
"logps/rejected": -608.5535888671875,
|
| 1122 |
+
"loss": 0.3074,
|
| 1123 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1124 |
+
"rewards/chosen": -1.422060251235962,
|
| 1125 |
+
"rewards/margins": 1.6324869394302368,
|
| 1126 |
+
"rewards/margins_max": 3.5064094066619873,
|
| 1127 |
+
"rewards/margins_min": 0.22324328124523163,
|
| 1128 |
+
"rewards/margins_std": 1.5147463083267212,
|
| 1129 |
+
"rewards/rejected": -3.054547071456909,
|
| 1130 |
+
"step": 560
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 1.61,
|
| 1134 |
+
"grad_norm": 4.771842727601967,
|
| 1135 |
+
"learning_rate": 2.6311127923312153e-07,
|
| 1136 |
+
"logits/chosen": -2.114218235015869,
|
| 1137 |
+
"logits/rejected": -1.990100622177124,
|
| 1138 |
+
"logps/chosen": -478.4190979003906,
|
| 1139 |
+
"logps/rejected": -668.0441284179688,
|
| 1140 |
+
"loss": 0.3081,
|
| 1141 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1142 |
+
"rewards/chosen": -1.6441261768341064,
|
| 1143 |
+
"rewards/margins": 1.6914678812026978,
|
| 1144 |
+
"rewards/margins_max": 3.585777759552002,
|
| 1145 |
+
"rewards/margins_min": 0.0474582239985466,
|
| 1146 |
+
"rewards/margins_std": 1.607506513595581,
|
| 1147 |
+
"rewards/rejected": -3.3355937004089355,
|
| 1148 |
+
"step": 570
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 1.63,
|
| 1152 |
+
"grad_norm": 4.718826798578273,
|
| 1153 |
+
"learning_rate": 2.5491866902400565e-07,
|
| 1154 |
+
"logits/chosen": -2.0991835594177246,
|
| 1155 |
+
"logits/rejected": -2.031989336013794,
|
| 1156 |
+
"logps/chosen": -415.6539611816406,
|
| 1157 |
+
"logps/rejected": -625.79443359375,
|
| 1158 |
+
"loss": 0.3326,
|
| 1159 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1160 |
+
"rewards/chosen": -1.5325905084609985,
|
| 1161 |
+
"rewards/margins": 1.6610767841339111,
|
| 1162 |
+
"rewards/margins_max": 3.7243950366973877,
|
| 1163 |
+
"rewards/margins_min": 0.19340097904205322,
|
| 1164 |
+
"rewards/margins_std": 1.592491865158081,
|
| 1165 |
+
"rewards/rejected": -3.19366717338562,
|
| 1166 |
+
"step": 580
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 1.66,
|
| 1170 |
+
"grad_norm": 6.197013861888915,
|
| 1171 |
+
"learning_rate": 2.4672076976812543e-07,
|
| 1172 |
+
"logits/chosen": -2.07092022895813,
|
| 1173 |
+
"logits/rejected": -1.9514614343643188,
|
| 1174 |
+
"logps/chosen": -456.1803283691406,
|
| 1175 |
+
"logps/rejected": -636.2240600585938,
|
| 1176 |
+
"loss": 0.3376,
|
| 1177 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1178 |
+
"rewards/chosen": -1.704035758972168,
|
| 1179 |
+
"rewards/margins": 1.6240657567977905,
|
| 1180 |
+
"rewards/margins_max": 3.5553812980651855,
|
| 1181 |
+
"rewards/margins_min": 0.13123703002929688,
|
| 1182 |
+
"rewards/margins_std": 1.570482850074768,
|
| 1183 |
+
"rewards/rejected": -3.328101396560669,
|
| 1184 |
+
"step": 590
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 1.69,
|
| 1188 |
+
"grad_norm": 4.544446788394471,
|
| 1189 |
+
"learning_rate": 2.385263966698222e-07,
|
| 1190 |
+
"logits/chosen": -2.005707263946533,
|
| 1191 |
+
"logits/rejected": -1.8865950107574463,
|
| 1192 |
+
"logps/chosen": -431.6078186035156,
|
| 1193 |
+
"logps/rejected": -682.1323852539062,
|
| 1194 |
+
"loss": 0.3449,
|
| 1195 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1196 |
+
"rewards/chosen": -1.613631248474121,
|
| 1197 |
+
"rewards/margins": 1.895965576171875,
|
| 1198 |
+
"rewards/margins_max": 4.0293354988098145,
|
| 1199 |
+
"rewards/margins_min": 0.33173805475234985,
|
| 1200 |
+
"rewards/margins_std": 1.6987276077270508,
|
| 1201 |
+
"rewards/rejected": -3.509597063064575,
|
| 1202 |
+
"step": 600
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 1.69,
|
| 1206 |
+
"eval_logits/chosen": -1.9706788063049316,
|
| 1207 |
+
"eval_logits/rejected": -1.882871389389038,
|
| 1208 |
+
"eval_logps/chosen": -505.41143798828125,
|
| 1209 |
+
"eval_logps/rejected": -519.8788452148438,
|
| 1210 |
+
"eval_loss": 0.8093447089195251,
|
| 1211 |
+
"eval_rewards/accuracies": 0.5730000138282776,
|
| 1212 |
+
"eval_rewards/chosen": -2.305307149887085,
|
| 1213 |
+
"eval_rewards/margins": 0.31827661395072937,
|
| 1214 |
+
"eval_rewards/margins_max": 2.791022539138794,
|
| 1215 |
+
"eval_rewards/margins_min": -1.984505295753479,
|
| 1216 |
+
"eval_rewards/margins_std": 1.5908182859420776,
|
| 1217 |
+
"eval_rewards/rejected": -2.6235837936401367,
|
| 1218 |
+
"eval_runtime": 740.929,
|
| 1219 |
+
"eval_samples_per_second": 2.699,
|
| 1220 |
+
"eval_steps_per_second": 0.169,
|
| 1221 |
+
"step": 600
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 1.72,
|
| 1225 |
+
"grad_norm": 7.884893832559347,
|
| 1226 |
+
"learning_rate": 2.3034436114175838e-07,
|
| 1227 |
+
"logits/chosen": -2.072605848312378,
|
| 1228 |
+
"logits/rejected": -1.8742202520370483,
|
| 1229 |
+
"logps/chosen": -470.3563537597656,
|
| 1230 |
+
"logps/rejected": -593.04345703125,
|
| 1231 |
+
"loss": 0.3553,
|
| 1232 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1233 |
+
"rewards/chosen": -1.6478776931762695,
|
| 1234 |
+
"rewards/margins": 1.5212364196777344,
|
| 1235 |
+
"rewards/margins_max": 3.331374406814575,
|
| 1236 |
+
"rewards/margins_min": 0.16380740702152252,
|
| 1237 |
+
"rewards/margins_std": 1.4223846197128296,
|
| 1238 |
+
"rewards/rejected": -3.169113874435425,
|
| 1239 |
+
"step": 610
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 1.75,
|
| 1243 |
+
"grad_norm": 4.6350166519672475,
|
| 1244 |
+
"learning_rate": 2.2218346133000264e-07,
|
| 1245 |
+
"logits/chosen": -2.145089864730835,
|
| 1246 |
+
"logits/rejected": -1.9883239269256592,
|
| 1247 |
+
"logps/chosen": -495.61676025390625,
|
| 1248 |
+
"logps/rejected": -693.98486328125,
|
| 1249 |
+
"loss": 0.3263,
|
| 1250 |
+
"rewards/accuracies": 0.875,
|
| 1251 |
+
"rewards/chosen": -1.746462106704712,
|
| 1252 |
+
"rewards/margins": 1.9258817434310913,
|
| 1253 |
+
"rewards/margins_max": 4.596188545227051,
|
| 1254 |
+
"rewards/margins_min": 0.21569041907787323,
|
| 1255 |
+
"rewards/margins_std": 2.0187416076660156,
|
| 1256 |
+
"rewards/rejected": -3.672344207763672,
|
| 1257 |
+
"step": 620
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 1.77,
|
| 1261 |
+
"grad_norm": 4.707195155665474,
|
| 1262 |
+
"learning_rate": 2.1405247265337917e-07,
|
| 1263 |
+
"logits/chosen": -2.0058627128601074,
|
| 1264 |
+
"logits/rejected": -1.8860944509506226,
|
| 1265 |
+
"logps/chosen": -489.9043884277344,
|
| 1266 |
+
"logps/rejected": -679.2588500976562,
|
| 1267 |
+
"loss": 0.3348,
|
| 1268 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1269 |
+
"rewards/chosen": -1.6168292760849,
|
| 1270 |
+
"rewards/margins": 1.956435203552246,
|
| 1271 |
+
"rewards/margins_max": 4.475503444671631,
|
| 1272 |
+
"rewards/margins_min": 0.15001405775547028,
|
| 1273 |
+
"rewards/margins_std": 1.965258002281189,
|
| 1274 |
+
"rewards/rejected": -3.5732643604278564,
|
| 1275 |
+
"step": 630
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 1.8,
|
| 1279 |
+
"grad_norm": 5.859575517402316,
|
| 1280 |
+
"learning_rate": 2.0596013836725657e-07,
|
| 1281 |
+
"logits/chosen": -2.161479949951172,
|
| 1282 |
+
"logits/rejected": -1.9951550960540771,
|
| 1283 |
+
"logps/chosen": -445.10345458984375,
|
| 1284 |
+
"logps/rejected": -593.2735595703125,
|
| 1285 |
+
"loss": 0.3148,
|
| 1286 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1287 |
+
"rewards/chosen": -1.5793910026550293,
|
| 1288 |
+
"rewards/margins": 1.6846816539764404,
|
| 1289 |
+
"rewards/margins_max": 3.323934555053711,
|
| 1290 |
+
"rewards/margins_min": 0.4667704701423645,
|
| 1291 |
+
"rewards/margins_std": 1.3031413555145264,
|
| 1292 |
+
"rewards/rejected": -3.264072895050049,
|
| 1293 |
+
"step": 640
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 1.83,
|
| 1297 |
+
"grad_norm": 7.1672923088518825,
|
| 1298 |
+
"learning_rate": 1.9791516016192213e-07,
|
| 1299 |
+
"logits/chosen": -2.07092547416687,
|
| 1300 |
+
"logits/rejected": -1.956458330154419,
|
| 1301 |
+
"logps/chosen": -485.3031311035156,
|
| 1302 |
+
"logps/rejected": -704.134765625,
|
| 1303 |
+
"loss": 0.3248,
|
| 1304 |
+
"rewards/accuracies": 0.875,
|
| 1305 |
+
"rewards/chosen": -1.7588361501693726,
|
| 1306 |
+
"rewards/margins": 1.9537765979766846,
|
| 1307 |
+
"rewards/margins_max": 4.279685020446777,
|
| 1308 |
+
"rewards/margins_min": 0.22622933983802795,
|
| 1309 |
+
"rewards/margins_std": 1.8460969924926758,
|
| 1310 |
+
"rewards/rejected": -3.7126126289367676,
|
| 1311 |
+
"step": 650
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 1.86,
|
| 1315 |
+
"grad_norm": 6.153621209141468,
|
| 1316 |
+
"learning_rate": 1.8992618880565036e-07,
|
| 1317 |
+
"logits/chosen": -2.0097153186798096,
|
| 1318 |
+
"logits/rejected": -1.9442319869995117,
|
| 1319 |
+
"logps/chosen": -366.12994384765625,
|
| 1320 |
+
"logps/rejected": -553.5570068359375,
|
| 1321 |
+
"loss": 0.3183,
|
| 1322 |
+
"rewards/accuracies": 0.875,
|
| 1323 |
+
"rewards/chosen": -1.3965609073638916,
|
| 1324 |
+
"rewards/margins": 1.6747844219207764,
|
| 1325 |
+
"rewards/margins_max": 3.8514389991760254,
|
| 1326 |
+
"rewards/margins_min": 0.07616128027439117,
|
| 1327 |
+
"rewards/margins_std": 1.7168182134628296,
|
| 1328 |
+
"rewards/rejected": -3.071345567703247,
|
| 1329 |
+
"step": 660
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 1.89,
|
| 1333 |
+
"grad_norm": 6.622934271289155,
|
| 1334 |
+
"learning_rate": 1.8200181484252885e-07,
|
| 1335 |
+
"logits/chosen": -2.1211202144622803,
|
| 1336 |
+
"logits/rejected": -2.00960373878479,
|
| 1337 |
+
"logps/chosen": -453.65386962890625,
|
| 1338 |
+
"logps/rejected": -652.305908203125,
|
| 1339 |
+
"loss": 0.3179,
|
| 1340 |
+
"rewards/accuracies": 0.875,
|
| 1341 |
+
"rewards/chosen": -1.7312113046646118,
|
| 1342 |
+
"rewards/margins": 2.003723621368408,
|
| 1343 |
+
"rewards/margins_max": 4.150836944580078,
|
| 1344 |
+
"rewards/margins_min": 0.22760996222496033,
|
| 1345 |
+
"rewards/margins_std": 1.777065634727478,
|
| 1346 |
+
"rewards/rejected": -3.7349350452423096,
|
| 1347 |
+
"step": 670
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 1.92,
|
| 1351 |
+
"grad_norm": 6.678835951751634,
|
| 1352 |
+
"learning_rate": 1.7415055935504233e-07,
|
| 1353 |
+
"logits/chosen": -2.175750970840454,
|
| 1354 |
+
"logits/rejected": -2.0145211219787598,
|
| 1355 |
+
"logps/chosen": -465.84259033203125,
|
| 1356 |
+
"logps/rejected": -653.8709106445312,
|
| 1357 |
+
"loss": 0.3322,
|
| 1358 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1359 |
+
"rewards/chosen": -1.7737252712249756,
|
| 1360 |
+
"rewards/margins": 1.8737843036651611,
|
| 1361 |
+
"rewards/margins_max": 4.312856197357178,
|
| 1362 |
+
"rewards/margins_min": -0.018575742840766907,
|
| 1363 |
+
"rewards/margins_std": 1.942803978919983,
|
| 1364 |
+
"rewards/rejected": -3.6475093364715576,
|
| 1365 |
+
"step": 680
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 1.94,
|
| 1369 |
+
"grad_norm": 5.060285672653984,
|
| 1370 |
+
"learning_rate": 1.6638086480134952e-07,
|
| 1371 |
+
"logits/chosen": -2.0581729412078857,
|
| 1372 |
+
"logits/rejected": -1.945577621459961,
|
| 1373 |
+
"logps/chosen": -358.5215148925781,
|
| 1374 |
+
"logps/rejected": -548.193359375,
|
| 1375 |
+
"loss": 0.3068,
|
| 1376 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1377 |
+
"rewards/chosen": -1.3307039737701416,
|
| 1378 |
+
"rewards/margins": 1.8142309188842773,
|
| 1379 |
+
"rewards/margins_max": 4.50165319442749,
|
| 1380 |
+
"rewards/margins_min": 0.17869335412979126,
|
| 1381 |
+
"rewards/margins_std": 1.988674521446228,
|
| 1382 |
+
"rewards/rejected": -3.144935131072998,
|
| 1383 |
+
"step": 690
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 1.97,
|
| 1387 |
+
"grad_norm": 8.639077018605542,
|
| 1388 |
+
"learning_rate": 1.5870108593710471e-07,
|
| 1389 |
+
"logits/chosen": -2.0774855613708496,
|
| 1390 |
+
"logits/rejected": -1.8586009740829468,
|
| 1391 |
+
"logps/chosen": -407.3385314941406,
|
| 1392 |
+
"logps/rejected": -548.2588500976562,
|
| 1393 |
+
"loss": 0.3253,
|
| 1394 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1395 |
+
"rewards/chosen": -1.3710554838180542,
|
| 1396 |
+
"rewards/margins": 1.697474718093872,
|
| 1397 |
+
"rewards/margins_max": 3.6990180015563965,
|
| 1398 |
+
"rewards/margins_min": 0.1754045933485031,
|
| 1399 |
+
"rewards/margins_std": 1.5944454669952393,
|
| 1400 |
+
"rewards/rejected": -3.0685300827026367,
|
| 1401 |
+
"step": 700
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 1.97,
|
| 1405 |
+
"eval_logits/chosen": -1.9524266719818115,
|
| 1406 |
+
"eval_logits/rejected": -1.8636794090270996,
|
| 1407 |
+
"eval_logps/chosen": -511.7565612792969,
|
| 1408 |
+
"eval_logps/rejected": -533.7400512695312,
|
| 1409 |
+
"eval_loss": 0.8021999001502991,
|
| 1410 |
+
"eval_rewards/accuracies": 0.5899999737739563,
|
| 1411 |
+
"eval_rewards/chosen": -2.3687589168548584,
|
| 1412 |
+
"eval_rewards/margins": 0.3934372663497925,
|
| 1413 |
+
"eval_rewards/margins_max": 3.059994697570801,
|
| 1414 |
+
"eval_rewards/margins_min": -2.04789400100708,
|
| 1415 |
+
"eval_rewards/margins_std": 1.6969114542007446,
|
| 1416 |
+
"eval_rewards/rejected": -2.7621960639953613,
|
| 1417 |
+
"eval_runtime": 740.07,
|
| 1418 |
+
"eval_samples_per_second": 2.702,
|
| 1419 |
+
"eval_steps_per_second": 0.169,
|
| 1420 |
+
"step": 700
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"epoch": 2.0,
|
| 1424 |
+
"grad_norm": 4.484030680446832,
|
| 1425 |
+
"learning_rate": 1.5111948083158528e-07,
|
| 1426 |
+
"logits/chosen": -1.9601835012435913,
|
| 1427 |
+
"logits/rejected": -1.8817923069000244,
|
| 1428 |
+
"logps/chosen": -376.6875915527344,
|
| 1429 |
+
"logps/rejected": -597.0089721679688,
|
| 1430 |
+
"loss": 0.2922,
|
| 1431 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1432 |
+
"rewards/chosen": -1.4828770160675049,
|
| 1433 |
+
"rewards/margins": 1.8921600580215454,
|
| 1434 |
+
"rewards/margins_max": 4.028073310852051,
|
| 1435 |
+
"rewards/margins_min": 0.3237634301185608,
|
| 1436 |
+
"rewards/margins_std": 1.6557111740112305,
|
| 1437 |
+
"rewards/rejected": -3.3750369548797607,
|
| 1438 |
+
"step": 710
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 2.03,
|
| 1442 |
+
"grad_norm": 5.934607513999122,
|
| 1443 |
+
"learning_rate": 1.4364420198778658e-07,
|
| 1444 |
+
"logits/chosen": -2.168855667114258,
|
| 1445 |
+
"logits/rejected": -1.9871803522109985,
|
| 1446 |
+
"logps/chosen": -517.0801391601562,
|
| 1447 |
+
"logps/rejected": -775.808349609375,
|
| 1448 |
+
"loss": 0.2773,
|
| 1449 |
+
"rewards/accuracies": 0.9375,
|
| 1450 |
+
"rewards/chosen": -1.8118972778320312,
|
| 1451 |
+
"rewards/margins": 2.484999656677246,
|
| 1452 |
+
"rewards/margins_max": 5.045625686645508,
|
| 1453 |
+
"rewards/margins_min": 0.6035622358322144,
|
| 1454 |
+
"rewards/margins_std": 2.0212478637695312,
|
| 1455 |
+
"rewards/rejected": -4.296896457672119,
|
| 1456 |
+
"step": 720
|
| 1457 |
+
},
|
| 1458 |
+
{
|
| 1459 |
+
"epoch": 2.06,
|
| 1460 |
+
"grad_norm": 6.664291467344353,
|
| 1461 |
+
"learning_rate": 1.3628328757603242e-07,
|
| 1462 |
+
"logits/chosen": -2.07096529006958,
|
| 1463 |
+
"logits/rejected": -1.88152277469635,
|
| 1464 |
+
"logps/chosen": -509.58056640625,
|
| 1465 |
+
"logps/rejected": -746.2061767578125,
|
| 1466 |
+
"loss": 0.2493,
|
| 1467 |
+
"rewards/accuracies": 0.9375,
|
| 1468 |
+
"rewards/chosen": -2.0079989433288574,
|
| 1469 |
+
"rewards/margins": 2.2790350914001465,
|
| 1470 |
+
"rewards/margins_max": 4.603484153747559,
|
| 1471 |
+
"rewards/margins_min": 0.4713754653930664,
|
| 1472 |
+
"rewards/margins_std": 1.8801381587982178,
|
| 1473 |
+
"rewards/rejected": -4.287034034729004,
|
| 1474 |
+
"step": 730
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 2.08,
|
| 1478 |
+
"grad_norm": 5.3728713322719175,
|
| 1479 |
+
"learning_rate": 1.2904465279052723e-07,
|
| 1480 |
+
"logits/chosen": -1.9976093769073486,
|
| 1481 |
+
"logits/rejected": -1.8471425771713257,
|
| 1482 |
+
"logps/chosen": -497.21484375,
|
| 1483 |
+
"logps/rejected": -705.7311401367188,
|
| 1484 |
+
"loss": 0.3031,
|
| 1485 |
+
"rewards/accuracies": 0.875,
|
| 1486 |
+
"rewards/chosen": -1.9563236236572266,
|
| 1487 |
+
"rewards/margins": 2.0360076427459717,
|
| 1488 |
+
"rewards/margins_max": 4.599600791931152,
|
| 1489 |
+
"rewards/margins_min": 0.06591819226741791,
|
| 1490 |
+
"rewards/margins_std": 2.065493583679199,
|
| 1491 |
+
"rewards/rejected": -3.9923312664031982,
|
| 1492 |
+
"step": 740
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 2.11,
|
| 1496 |
+
"grad_norm": 4.3979433368191625,
|
| 1497 |
+
"learning_rate": 1.219360813381446e-07,
|
| 1498 |
+
"logits/chosen": -1.9922767877578735,
|
| 1499 |
+
"logits/rejected": -1.8619979619979858,
|
| 1500 |
+
"logps/chosen": -434.5537109375,
|
| 1501 |
+
"logps/rejected": -657.9417724609375,
|
| 1502 |
+
"loss": 0.288,
|
| 1503 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1504 |
+
"rewards/chosen": -1.721703290939331,
|
| 1505 |
+
"rewards/margins": 2.172398090362549,
|
| 1506 |
+
"rewards/margins_max": 4.379127025604248,
|
| 1507 |
+
"rewards/margins_min": 0.2665182948112488,
|
| 1508 |
+
"rewards/margins_std": 1.8405358791351318,
|
| 1509 |
+
"rewards/rejected": -3.8941009044647217,
|
| 1510 |
+
"step": 750
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 2.14,
|
| 1514 |
+
"grad_norm": 4.311152372020765,
|
| 1515 |
+
"learning_rate": 1.149652170686039e-07,
|
| 1516 |
+
"logits/chosen": -1.910975456237793,
|
| 1517 |
+
"logits/rejected": -1.7928791046142578,
|
| 1518 |
+
"logps/chosen": -441.78240966796875,
|
| 1519 |
+
"logps/rejected": -715.9662475585938,
|
| 1520 |
+
"loss": 0.2896,
|
| 1521 |
+
"rewards/accuracies": 0.9375,
|
| 1522 |
+
"rewards/chosen": -1.777691125869751,
|
| 1523 |
+
"rewards/margins": 2.4173693656921387,
|
| 1524 |
+
"rewards/margins_max": 5.414074897766113,
|
| 1525 |
+
"rewards/margins_min": 0.11628694832324982,
|
| 1526 |
+
"rewards/margins_std": 2.435407876968384,
|
| 1527 |
+
"rewards/rejected": -4.195060729980469,
|
| 1528 |
+
"step": 760
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 2.17,
|
| 1532 |
+
"grad_norm": 4.632824439408098,
|
| 1533 |
+
"learning_rate": 1.0813955575503587e-07,
|
| 1534 |
+
"logits/chosen": -2.0070652961730957,
|
| 1535 |
+
"logits/rejected": -1.8251206874847412,
|
| 1536 |
+
"logps/chosen": -446.7857971191406,
|
| 1537 |
+
"logps/rejected": -642.5098876953125,
|
| 1538 |
+
"loss": 0.2992,
|
| 1539 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1540 |
+
"rewards/chosen": -1.7226359844207764,
|
| 1541 |
+
"rewards/margins": 2.2547175884246826,
|
| 1542 |
+
"rewards/margins_max": 5.0700860023498535,
|
| 1543 |
+
"rewards/margins_min": 0.39999958872795105,
|
| 1544 |
+
"rewards/margins_std": 2.0922274589538574,
|
| 1545 |
+
"rewards/rejected": -3.977353572845459,
|
| 1546 |
+
"step": 770
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 2.2,
|
| 1550 |
+
"grad_norm": 7.187846299826648,
|
| 1551 |
+
"learning_rate": 1.0146643703377486e-07,
|
| 1552 |
+
"logits/chosen": -2.025348424911499,
|
| 1553 |
+
"logits/rejected": -1.8247390985488892,
|
| 1554 |
+
"logps/chosen": -516.863525390625,
|
| 1555 |
+
"logps/rejected": -660.7418212890625,
|
| 1556 |
+
"loss": 0.3685,
|
| 1557 |
+
"rewards/accuracies": 0.8125,
|
| 1558 |
+
"rewards/chosen": -2.216492176055908,
|
| 1559 |
+
"rewards/margins": 1.8119462728500366,
|
| 1560 |
+
"rewards/margins_max": 4.643686294555664,
|
| 1561 |
+
"rewards/margins_min": -0.28089627623558044,
|
| 1562 |
+
"rewards/margins_std": 2.2221999168395996,
|
| 1563 |
+
"rewards/rejected": -4.028437614440918,
|
| 1564 |
+
"step": 780
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 2.23,
|
| 1568 |
+
"grad_norm": 4.864775866905843,
|
| 1569 |
+
"learning_rate": 9.495303651204494e-08,
|
| 1570 |
+
"logits/chosen": -2.032228946685791,
|
| 1571 |
+
"logits/rejected": -1.8665939569473267,
|
| 1572 |
+
"logps/chosen": -472.88018798828125,
|
| 1573 |
+
"logps/rejected": -657.1442260742188,
|
| 1574 |
+
"loss": 0.2754,
|
| 1575 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1576 |
+
"rewards/chosen": -1.7393814325332642,
|
| 1577 |
+
"rewards/margins": 1.9806731939315796,
|
| 1578 |
+
"rewards/margins_max": 4.059412002563477,
|
| 1579 |
+
"rewards/margins_min": 0.27446281909942627,
|
| 1580 |
+
"rewards/margins_std": 1.732347846031189,
|
| 1581 |
+
"rewards/rejected": -3.7200546264648438,
|
| 1582 |
+
"step": 790
|
| 1583 |
+
},
|
| 1584 |
+
{
|
| 1585 |
+
"epoch": 2.25,
|
| 1586 |
+
"grad_norm": 4.897791088230011,
|
| 1587 |
+
"learning_rate": 8.860635805202615e-08,
|
| 1588 |
+
"logits/chosen": -2.0725150108337402,
|
| 1589 |
+
"logits/rejected": -1.939500093460083,
|
| 1590 |
+
"logps/chosen": -462.3565368652344,
|
| 1591 |
+
"logps/rejected": -782.6009521484375,
|
| 1592 |
+
"loss": 0.2445,
|
| 1593 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 1594 |
+
"rewards/chosen": -1.6571842432022095,
|
| 1595 |
+
"rewards/margins": 2.6175780296325684,
|
| 1596 |
+
"rewards/margins_max": 5.12880802154541,
|
| 1597 |
+
"rewards/margins_min": 0.6253465414047241,
|
| 1598 |
+
"rewards/margins_std": 2.0294041633605957,
|
| 1599 |
+
"rewards/rejected": -4.274762153625488,
|
| 1600 |
+
"step": 800
|
| 1601 |
+
},
|
| 1602 |
+
{
|
| 1603 |
+
"epoch": 2.25,
|
| 1604 |
+
"eval_logits/chosen": -1.919421672821045,
|
| 1605 |
+
"eval_logits/rejected": -1.8329098224639893,
|
| 1606 |
+
"eval_logps/chosen": -536.6691284179688,
|
| 1607 |
+
"eval_logps/rejected": -563.362060546875,
|
| 1608 |
+
"eval_loss": 0.8262488842010498,
|
| 1609 |
+
"eval_rewards/accuracies": 0.5879999995231628,
|
| 1610 |
+
"eval_rewards/chosen": -2.6178839206695557,
|
| 1611 |
+
"eval_rewards/margins": 0.44053173065185547,
|
| 1612 |
+
"eval_rewards/margins_max": 3.385202407836914,
|
| 1613 |
+
"eval_rewards/margins_min": -2.237797260284424,
|
| 1614 |
+
"eval_rewards/margins_std": 1.8657594919204712,
|
| 1615 |
+
"eval_rewards/rejected": -3.0584161281585693,
|
| 1616 |
+
"eval_runtime": 740.1616,
|
| 1617 |
+
"eval_samples_per_second": 2.702,
|
| 1618 |
+
"eval_steps_per_second": 0.169,
|
| 1619 |
+
"step": 800
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 2.28,
|
| 1623 |
+
"grad_norm": 3.8140500779127393,
|
| 1624 |
+
"learning_rate": 8.24332262395994e-08,
|
| 1625 |
+
"logits/chosen": -2.062171459197998,
|
| 1626 |
+
"logits/rejected": -1.9682680368423462,
|
| 1627 |
+
"logps/chosen": -468.1681213378906,
|
| 1628 |
+
"logps/rejected": -737.950439453125,
|
| 1629 |
+
"loss": 0.2607,
|
| 1630 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1631 |
+
"rewards/chosen": -1.9337375164031982,
|
| 1632 |
+
"rewards/margins": 2.346217393875122,
|
| 1633 |
+
"rewards/margins_max": 4.732678413391113,
|
| 1634 |
+
"rewards/margins_min": 0.3489117920398712,
|
| 1635 |
+
"rewards/margins_std": 1.9971859455108643,
|
| 1636 |
+
"rewards/rejected": -4.2799553871154785,
|
| 1637 |
+
"step": 810
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 2.31,
|
| 1641 |
+
"grad_norm": 4.503876817512786,
|
| 1642 |
+
"learning_rate": 7.644027904586586e-08,
|
| 1643 |
+
"logits/chosen": -1.9747568368911743,
|
| 1644 |
+
"logits/rejected": -1.861803412437439,
|
| 1645 |
+
"logps/chosen": -466.3984375,
|
| 1646 |
+
"logps/rejected": -755.3158569335938,
|
| 1647 |
+
"loss": 0.26,
|
| 1648 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1649 |
+
"rewards/chosen": -1.7740318775177002,
|
| 1650 |
+
"rewards/margins": 2.602890968322754,
|
| 1651 |
+
"rewards/margins_max": 5.653742790222168,
|
| 1652 |
+
"rewards/margins_min": 0.4451362192630768,
|
| 1653 |
+
"rewards/margins_std": 2.3605756759643555,
|
| 1654 |
+
"rewards/rejected": -4.376922607421875,
|
| 1655 |
+
"step": 820
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 2.34,
|
| 1659 |
+
"grad_norm": 8.38717540675709,
|
| 1660 |
+
"learning_rate": 7.063396068933469e-08,
|
| 1661 |
+
"logits/chosen": -1.9597285985946655,
|
| 1662 |
+
"logits/rejected": -1.7506755590438843,
|
| 1663 |
+
"logps/chosen": -575.2877197265625,
|
| 1664 |
+
"logps/rejected": -752.0540161132812,
|
| 1665 |
+
"loss": 0.289,
|
| 1666 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1667 |
+
"rewards/chosen": -2.2225418090820312,
|
| 1668 |
+
"rewards/margins": 2.1929917335510254,
|
| 1669 |
+
"rewards/margins_max": 4.804136753082275,
|
| 1670 |
+
"rewards/margins_min": 0.21819576621055603,
|
| 1671 |
+
"rewards/margins_std": 2.0718307495117188,
|
| 1672 |
+
"rewards/rejected": -4.415533542633057,
|
| 1673 |
+
"step": 830
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 2.37,
|
| 1677 |
+
"grad_norm": 4.269821171109155,
|
| 1678 |
+
"learning_rate": 6.502051470645148e-08,
|
| 1679 |
+
"logits/chosen": -2.0234594345092773,
|
| 1680 |
+
"logits/rejected": -1.8737064599990845,
|
| 1681 |
+
"logps/chosen": -497.57843017578125,
|
| 1682 |
+
"logps/rejected": -664.8786010742188,
|
| 1683 |
+
"loss": 0.2876,
|
| 1684 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1685 |
+
"rewards/chosen": -1.9449161291122437,
|
| 1686 |
+
"rewards/margins": 1.8489919900894165,
|
| 1687 |
+
"rewards/margins_max": 3.4735729694366455,
|
| 1688 |
+
"rewards/margins_min": 0.39146485924720764,
|
| 1689 |
+
"rewards/margins_std": 1.407354712486267,
|
| 1690 |
+
"rewards/rejected": -3.7939083576202393,
|
| 1691 |
+
"step": 840
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 2.39,
|
| 1695 |
+
"grad_norm": 6.881439522360299,
|
| 1696 |
+
"learning_rate": 5.960597723792194e-08,
|
| 1697 |
+
"logits/chosen": -2.020962715148926,
|
| 1698 |
+
"logits/rejected": -1.8502124547958374,
|
| 1699 |
+
"logps/chosen": -445.7184143066406,
|
| 1700 |
+
"logps/rejected": -693.6365356445312,
|
| 1701 |
+
"loss": 0.2767,
|
| 1702 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1703 |
+
"rewards/chosen": -1.7460823059082031,
|
| 1704 |
+
"rewards/margins": 2.4407520294189453,
|
| 1705 |
+
"rewards/margins_max": 4.710750102996826,
|
| 1706 |
+
"rewards/margins_min": 0.3442041873931885,
|
| 1707 |
+
"rewards/margins_std": 1.993112564086914,
|
| 1708 |
+
"rewards/rejected": -4.18683385848999,
|
| 1709 |
+
"step": 850
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 2.42,
|
| 1713 |
+
"grad_norm": 7.845678328004129,
|
| 1714 |
+
"learning_rate": 5.4396170538046486e-08,
|
| 1715 |
+
"logits/chosen": -2.072786808013916,
|
| 1716 |
+
"logits/rejected": -1.908163070678711,
|
| 1717 |
+
"logps/chosen": -511.01104736328125,
|
| 1718 |
+
"logps/rejected": -751.96435546875,
|
| 1719 |
+
"loss": 0.2501,
|
| 1720 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1721 |
+
"rewards/chosen": -1.9439871311187744,
|
| 1722 |
+
"rewards/margins": 2.6217305660247803,
|
| 1723 |
+
"rewards/margins_max": 5.660550594329834,
|
| 1724 |
+
"rewards/margins_min": 0.2680825889110565,
|
| 1725 |
+
"rewards/margins_std": 2.4432153701782227,
|
| 1726 |
+
"rewards/rejected": -4.565717697143555,
|
| 1727 |
+
"step": 860
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 2.45,
|
| 1731 |
+
"grad_norm": 7.4391577660258195,
|
| 1732 |
+
"learning_rate": 4.93966967140487e-08,
|
| 1733 |
+
"logits/chosen": -1.9789683818817139,
|
| 1734 |
+
"logits/rejected": -1.842271089553833,
|
| 1735 |
+
"logps/chosen": -460.94091796875,
|
| 1736 |
+
"logps/rejected": -734.498779296875,
|
| 1737 |
+
"loss": 0.2516,
|
| 1738 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1739 |
+
"rewards/chosen": -1.7545455694198608,
|
| 1740 |
+
"rewards/margins": 2.402285575866699,
|
| 1741 |
+
"rewards/margins_max": 5.085206985473633,
|
| 1742 |
+
"rewards/margins_min": 0.28181013464927673,
|
| 1743 |
+
"rewards/margins_std": 2.1507792472839355,
|
| 1744 |
+
"rewards/rejected": -4.156831741333008,
|
| 1745 |
+
"step": 870
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 2.48,
|
| 1749 |
+
"grad_norm": 5.333353916764276,
|
| 1750 |
+
"learning_rate": 4.4612931702126433e-08,
|
| 1751 |
+
"logits/chosen": -2.1065726280212402,
|
| 1752 |
+
"logits/rejected": -1.9196290969848633,
|
| 1753 |
+
"logps/chosen": -523.8679809570312,
|
| 1754 |
+
"logps/rejected": -738.0938720703125,
|
| 1755 |
+
"loss": 0.2638,
|
| 1756 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1757 |
+
"rewards/chosen": -2.153127908706665,
|
| 1758 |
+
"rewards/margins": 2.2366292476654053,
|
| 1759 |
+
"rewards/margins_max": 4.850363254547119,
|
| 1760 |
+
"rewards/margins_min": 0.37360453605651855,
|
| 1761 |
+
"rewards/margins_std": 2.033721923828125,
|
| 1762 |
+
"rewards/rejected": -4.38975715637207,
|
| 1763 |
+
"step": 880
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 2.51,
|
| 1767 |
+
"grad_norm": 5.085291105056715,
|
| 1768 |
+
"learning_rate": 4.005001948670605e-08,
|
| 1769 |
+
"logits/chosen": -2.0524444580078125,
|
| 1770 |
+
"logits/rejected": -1.8745992183685303,
|
| 1771 |
+
"logps/chosen": -531.6593017578125,
|
| 1772 |
+
"logps/rejected": -733.5911865234375,
|
| 1773 |
+
"loss": 0.2719,
|
| 1774 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1775 |
+
"rewards/chosen": -2.118652820587158,
|
| 1776 |
+
"rewards/margins": 2.1900432109832764,
|
| 1777 |
+
"rewards/margins_max": 4.532656669616699,
|
| 1778 |
+
"rewards/margins_min": 0.4483065605163574,
|
| 1779 |
+
"rewards/margins_std": 1.870699167251587,
|
| 1780 |
+
"rewards/rejected": -4.3086957931518555,
|
| 1781 |
+
"step": 890
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 2.54,
|
| 1785 |
+
"grad_norm": 6.4536898091014265,
|
| 1786 |
+
"learning_rate": 3.571286656911376e-08,
|
| 1787 |
+
"logits/chosen": -2.048665761947632,
|
| 1788 |
+
"logits/rejected": -1.8375803232192993,
|
| 1789 |
+
"logps/chosen": -514.3804321289062,
|
| 1790 |
+
"logps/rejected": -805.6507568359375,
|
| 1791 |
+
"loss": 0.3015,
|
| 1792 |
+
"rewards/accuracies": 0.9375,
|
| 1793 |
+
"rewards/chosen": -2.077589750289917,
|
| 1794 |
+
"rewards/margins": 2.7033679485321045,
|
| 1795 |
+
"rewards/margins_max": 5.94973087310791,
|
| 1796 |
+
"rewards/margins_min": 0.318843811750412,
|
| 1797 |
+
"rewards/margins_std": 2.5493154525756836,
|
| 1798 |
+
"rewards/rejected": -4.7809576988220215,
|
| 1799 |
+
"step": 900
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 2.54,
|
| 1803 |
+
"eval_logits/chosen": -1.9146674871444702,
|
| 1804 |
+
"eval_logits/rejected": -1.8280622959136963,
|
| 1805 |
+
"eval_logps/chosen": -542.6185302734375,
|
| 1806 |
+
"eval_logps/rejected": -571.6796264648438,
|
| 1807 |
+
"eval_loss": 0.8292613625526428,
|
| 1808 |
+
"eval_rewards/accuracies": 0.5929999947547913,
|
| 1809 |
+
"eval_rewards/chosen": -2.6773781776428223,
|
| 1810 |
+
"eval_rewards/margins": 0.46421319246292114,
|
| 1811 |
+
"eval_rewards/margins_max": 3.504265308380127,
|
| 1812 |
+
"eval_rewards/margins_min": -2.2912440299987793,
|
| 1813 |
+
"eval_rewards/margins_std": 1.918405294418335,
|
| 1814 |
+
"eval_rewards/rejected": -3.1415910720825195,
|
| 1815 |
+
"eval_runtime": 740.6009,
|
| 1816 |
+
"eval_samples_per_second": 2.701,
|
| 1817 |
+
"eval_steps_per_second": 0.169,
|
| 1818 |
+
"step": 900
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 2.56,
|
| 1822 |
+
"grad_norm": 4.3310162633935345,
|
| 1823 |
+
"learning_rate": 3.160613669161255e-08,
|
| 1824 |
+
"logits/chosen": -2.009597063064575,
|
| 1825 |
+
"logits/rejected": -1.8357868194580078,
|
| 1826 |
+
"logps/chosen": -472.4285583496094,
|
| 1827 |
+
"logps/rejected": -671.4129028320312,
|
| 1828 |
+
"loss": 0.2756,
|
| 1829 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1830 |
+
"rewards/chosen": -1.863695502281189,
|
| 1831 |
+
"rewards/margins": 2.203331232070923,
|
| 1832 |
+
"rewards/margins_max": 4.663190841674805,
|
| 1833 |
+
"rewards/margins_min": 0.4688544273376465,
|
| 1834 |
+
"rewards/margins_std": 1.9418967962265015,
|
| 1835 |
+
"rewards/rejected": -4.067026615142822,
|
| 1836 |
+
"step": 910
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 2.59,
|
| 1840 |
+
"grad_norm": 5.566695099413474,
|
| 1841 |
+
"learning_rate": 2.7734245822478436e-08,
|
| 1842 |
+
"logits/chosen": -2.015672206878662,
|
| 1843 |
+
"logits/rejected": -1.8013242483139038,
|
| 1844 |
+
"logps/chosen": -464.11163330078125,
|
| 1845 |
+
"logps/rejected": -615.1932373046875,
|
| 1846 |
+
"loss": 0.3132,
|
| 1847 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1848 |
+
"rewards/chosen": -1.7841193675994873,
|
| 1849 |
+
"rewards/margins": 1.9425182342529297,
|
| 1850 |
+
"rewards/margins_max": 4.301909923553467,
|
| 1851 |
+
"rewards/margins_min": 0.10777553170919418,
|
| 1852 |
+
"rewards/margins_std": 1.9170602560043335,
|
| 1853 |
+
"rewards/rejected": -3.726637363433838,
|
| 1854 |
+
"step": 920
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 2.62,
|
| 1858 |
+
"grad_norm": 7.3325072892546554,
|
| 1859 |
+
"learning_rate": 2.410135740750821e-08,
|
| 1860 |
+
"logits/chosen": -2.0186896324157715,
|
| 1861 |
+
"logits/rejected": -1.8880863189697266,
|
| 1862 |
+
"logps/chosen": -460.39154052734375,
|
| 1863 |
+
"logps/rejected": -668.5147705078125,
|
| 1864 |
+
"loss": 0.2969,
|
| 1865 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1866 |
+
"rewards/chosen": -1.975254774093628,
|
| 1867 |
+
"rewards/margins": 2.0803561210632324,
|
| 1868 |
+
"rewards/margins_max": 4.375603675842285,
|
| 1869 |
+
"rewards/margins_min": 0.5560076236724854,
|
| 1870 |
+
"rewards/margins_std": 1.7459514141082764,
|
| 1871 |
+
"rewards/rejected": -4.055610656738281,
|
| 1872 |
+
"step": 930
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 2.65,
|
| 1876 |
+
"grad_norm": 4.1844081566816715,
|
| 1877 |
+
"learning_rate": 2.071137789306418e-08,
|
| 1878 |
+
"logits/chosen": -2.073499917984009,
|
| 1879 |
+
"logits/rejected": -1.8808135986328125,
|
| 1880 |
+
"logps/chosen": -530.9814453125,
|
| 1881 |
+
"logps/rejected": -763.3355712890625,
|
| 1882 |
+
"loss": 0.2802,
|
| 1883 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1884 |
+
"rewards/chosen": -2.0685553550720215,
|
| 1885 |
+
"rewards/margins": 2.229001998901367,
|
| 1886 |
+
"rewards/margins_max": 5.061312198638916,
|
| 1887 |
+
"rewards/margins_min": 0.28276023268699646,
|
| 1888 |
+
"rewards/margins_std": 2.12688946723938,
|
| 1889 |
+
"rewards/rejected": -4.297557353973389,
|
| 1890 |
+
"step": 940
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 2.68,
|
| 1894 |
+
"grad_norm": 5.379046668918782,
|
| 1895 |
+
"learning_rate": 1.7567952525471107e-08,
|
| 1896 |
+
"logits/chosen": -1.981681227684021,
|
| 1897 |
+
"logits/rejected": -1.835107445716858,
|
| 1898 |
+
"logps/chosen": -423.8411560058594,
|
| 1899 |
+
"logps/rejected": -588.3886108398438,
|
| 1900 |
+
"loss": 0.3094,
|
| 1901 |
+
"rewards/accuracies": 0.9375,
|
| 1902 |
+
"rewards/chosen": -1.7472492456436157,
|
| 1903 |
+
"rewards/margins": 1.7251592874526978,
|
| 1904 |
+
"rewards/margins_max": 3.747589111328125,
|
| 1905 |
+
"rewards/margins_min": 0.21285438537597656,
|
| 1906 |
+
"rewards/margins_std": 1.5877231359481812,
|
| 1907 |
+
"rewards/rejected": -3.4724082946777344,
|
| 1908 |
+
"step": 950
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 2.7,
|
| 1912 |
+
"grad_norm": 5.323826462643735,
|
| 1913 |
+
"learning_rate": 1.467446143128101e-08,
|
| 1914 |
+
"logits/chosen": -2.054689884185791,
|
| 1915 |
+
"logits/rejected": -1.922407865524292,
|
| 1916 |
+
"logps/chosen": -434.65081787109375,
|
| 1917 |
+
"logps/rejected": -617.8987426757812,
|
| 1918 |
+
"loss": 0.2779,
|
| 1919 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1920 |
+
"rewards/chosen": -1.7226186990737915,
|
| 1921 |
+
"rewards/margins": 1.9324982166290283,
|
| 1922 |
+
"rewards/margins_max": 3.994910478591919,
|
| 1923 |
+
"rewards/margins_min": 0.3764217495918274,
|
| 1924 |
+
"rewards/margins_std": 1.6352239847183228,
|
| 1925 |
+
"rewards/rejected": -3.6551170349121094,
|
| 1926 |
+
"step": 960
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 2.73,
|
| 1930 |
+
"grad_norm": 9.489539963211016,
|
| 1931 |
+
"learning_rate": 1.2034015982622243e-08,
|
| 1932 |
+
"logits/chosen": -2.103243112564087,
|
| 1933 |
+
"logits/rejected": -1.8488214015960693,
|
| 1934 |
+
"logps/chosen": -522.6852416992188,
|
| 1935 |
+
"logps/rejected": -773.9171752929688,
|
| 1936 |
+
"loss": 0.2864,
|
| 1937 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1938 |
+
"rewards/chosen": -1.9787250757217407,
|
| 1939 |
+
"rewards/margins": 2.53349232673645,
|
| 1940 |
+
"rewards/margins_max": 5.404959678649902,
|
| 1941 |
+
"rewards/margins_min": 0.26877671480178833,
|
| 1942 |
+
"rewards/margins_std": 2.34672212600708,
|
| 1943 |
+
"rewards/rejected": -4.5122175216674805,
|
| 1944 |
+
"step": 970
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 2.76,
|
| 1948 |
+
"grad_norm": 4.819958146732696,
|
| 1949 |
+
"learning_rate": 9.649455451539418e-09,
|
| 1950 |
+
"logits/chosen": -2.0097153186798096,
|
| 1951 |
+
"logits/rejected": -1.8648490905761719,
|
| 1952 |
+
"logps/chosen": -453.50042724609375,
|
| 1953 |
+
"logps/rejected": -675.0182495117188,
|
| 1954 |
+
"loss": 0.2579,
|
| 1955 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1956 |
+
"rewards/chosen": -1.9288856983184814,
|
| 1957 |
+
"rewards/margins": 2.2360923290252686,
|
| 1958 |
+
"rewards/margins_max": 4.845951080322266,
|
| 1959 |
+
"rewards/margins_min": 0.353663831949234,
|
| 1960 |
+
"rewards/margins_std": 2.0667340755462646,
|
| 1961 |
+
"rewards/rejected": -4.16497802734375,
|
| 1962 |
+
"step": 980
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 2.79,
|
| 1966 |
+
"grad_norm": 4.860055955478379,
|
| 1967 |
+
"learning_rate": 7.523343956923194e-09,
|
| 1968 |
+
"logits/chosen": -1.9947786331176758,
|
| 1969 |
+
"logits/rejected": -1.7985414266586304,
|
| 1970 |
+
"logps/chosen": -480.50103759765625,
|
| 1971 |
+
"logps/rejected": -720.1832885742188,
|
| 1972 |
+
"loss": 0.307,
|
| 1973 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1974 |
+
"rewards/chosen": -1.8668218851089478,
|
| 1975 |
+
"rewards/margins": 2.3755202293395996,
|
| 1976 |
+
"rewards/margins_max": 5.254766941070557,
|
| 1977 |
+
"rewards/margins_min": 0.25941935181617737,
|
| 1978 |
+
"rewards/margins_std": 2.237616777420044,
|
| 1979 |
+
"rewards/rejected": -4.242341995239258,
|
| 1980 |
+
"step": 990
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 2.82,
|
| 1984 |
+
"grad_norm": 4.887859891241385,
|
| 1985 |
+
"learning_rate": 5.6579677073121945e-09,
|
| 1986 |
+
"logits/chosen": -1.95950448513031,
|
| 1987 |
+
"logits/rejected": -1.8909938335418701,
|
| 1988 |
+
"logps/chosen": -456.54913330078125,
|
| 1989 |
+
"logps/rejected": -741.733642578125,
|
| 1990 |
+
"loss": 0.2725,
|
| 1991 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1992 |
+
"rewards/chosen": -1.8867641687393188,
|
| 1993 |
+
"rewards/margins": 2.411424160003662,
|
| 1994 |
+
"rewards/margins_max": 5.223512649536133,
|
| 1995 |
+
"rewards/margins_min": 0.1520969420671463,
|
| 1996 |
+
"rewards/margins_std": 2.2634613513946533,
|
| 1997 |
+
"rewards/rejected": -4.29818868637085,
|
| 1998 |
+
"step": 1000
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 2.82,
|
| 2002 |
+
"eval_logits/chosen": -1.9147663116455078,
|
| 2003 |
+
"eval_logits/rejected": -1.8277100324630737,
|
| 2004 |
+
"eval_logps/chosen": -539.9697265625,
|
| 2005 |
+
"eval_logps/rejected": -569.4471435546875,
|
| 2006 |
+
"eval_loss": 0.825098991394043,
|
| 2007 |
+
"eval_rewards/accuracies": 0.5929999947547913,
|
| 2008 |
+
"eval_rewards/chosen": -2.6508901119232178,
|
| 2009 |
+
"eval_rewards/margins": 0.4683758020401001,
|
| 2010 |
+
"eval_rewards/margins_max": 3.500056505203247,
|
| 2011 |
+
"eval_rewards/margins_min": -2.2740743160247803,
|
| 2012 |
+
"eval_rewards/margins_std": 1.9113789796829224,
|
| 2013 |
+
"eval_rewards/rejected": -3.1192662715911865,
|
| 2014 |
+
"eval_runtime": 741.0299,
|
| 2015 |
+
"eval_samples_per_second": 2.699,
|
| 2016 |
+
"eval_steps_per_second": 0.169,
|
| 2017 |
+
"step": 1000
|
| 2018 |
+
},
|
| 2019 |
+
{
|
| 2020 |
+
"epoch": 2.85,
|
| 2021 |
+
"grad_norm": 6.523826359106252,
|
| 2022 |
+
"learning_rate": 4.0553325425319585e-09,
|
| 2023 |
+
"logits/chosen": -2.071352243423462,
|
| 2024 |
+
"logits/rejected": -1.9273669719696045,
|
| 2025 |
+
"logps/chosen": -521.459228515625,
|
| 2026 |
+
"logps/rejected": -810.1173095703125,
|
| 2027 |
+
"loss": 0.3486,
|
| 2028 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 2029 |
+
"rewards/chosen": -2.1784815788269043,
|
| 2030 |
+
"rewards/margins": 2.454408645629883,
|
| 2031 |
+
"rewards/margins_max": 5.279435157775879,
|
| 2032 |
+
"rewards/margins_min": 0.46752485632896423,
|
| 2033 |
+
"rewards/margins_std": 2.169618844985962,
|
| 2034 |
+
"rewards/rejected": -4.632889747619629,
|
| 2035 |
+
"step": 1010
|
| 2036 |
+
},
|
| 2037 |
+
{
|
| 2038 |
+
"epoch": 2.87,
|
| 2039 |
+
"grad_norm": 9.128100978737436,
|
| 2040 |
+
"learning_rate": 2.717161776814747e-09,
|
| 2041 |
+
"logits/chosen": -2.0809285640716553,
|
| 2042 |
+
"logits/rejected": -1.9295809268951416,
|
| 2043 |
+
"logps/chosen": -497.36962890625,
|
| 2044 |
+
"logps/rejected": -737.8751831054688,
|
| 2045 |
+
"loss": 0.284,
|
| 2046 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2047 |
+
"rewards/chosen": -2.0471560955047607,
|
| 2048 |
+
"rewards/margins": 2.289663791656494,
|
| 2049 |
+
"rewards/margins_max": 5.025771141052246,
|
| 2050 |
+
"rewards/margins_min": 0.4177941679954529,
|
| 2051 |
+
"rewards/margins_std": 2.142401695251465,
|
| 2052 |
+
"rewards/rejected": -4.336819648742676,
|
| 2053 |
+
"step": 1020
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 2.9,
|
| 2057 |
+
"grad_norm": 7.711600386273794,
|
| 2058 |
+
"learning_rate": 1.6448943457189613e-09,
|
| 2059 |
+
"logits/chosen": -2.0054070949554443,
|
| 2060 |
+
"logits/rejected": -1.8348314762115479,
|
| 2061 |
+
"logps/chosen": -463.53558349609375,
|
| 2062 |
+
"logps/rejected": -709.8609619140625,
|
| 2063 |
+
"loss": 0.3101,
|
| 2064 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2065 |
+
"rewards/chosen": -1.7694952487945557,
|
| 2066 |
+
"rewards/margins": 2.3598575592041016,
|
| 2067 |
+
"rewards/margins_max": 5.3210344314575195,
|
| 2068 |
+
"rewards/margins_min": 0.4152081608772278,
|
| 2069 |
+
"rewards/margins_std": 2.2792410850524902,
|
| 2070 |
+
"rewards/rejected": -4.129352569580078,
|
| 2071 |
+
"step": 1030
|
| 2072 |
+
},
|
| 2073 |
+
{
|
| 2074 |
+
"epoch": 2.93,
|
| 2075 |
+
"grad_norm": 7.482361970769731,
|
| 2076 |
+
"learning_rate": 8.396832588411229e-10,
|
| 2077 |
+
"logits/chosen": -1.911126732826233,
|
| 2078 |
+
"logits/rejected": -1.759119987487793,
|
| 2079 |
+
"logps/chosen": -461.5048828125,
|
| 2080 |
+
"logps/rejected": -633.4130859375,
|
| 2081 |
+
"loss": 0.3082,
|
| 2082 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 2083 |
+
"rewards/chosen": -1.928013563156128,
|
| 2084 |
+
"rewards/margins": 1.7119224071502686,
|
| 2085 |
+
"rewards/margins_max": 4.119196891784668,
|
| 2086 |
+
"rewards/margins_min": 0.11787784099578857,
|
| 2087 |
+
"rewards/margins_std": 1.8589006662368774,
|
| 2088 |
+
"rewards/rejected": -3.6399359703063965,
|
| 2089 |
+
"step": 1040
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 2.96,
|
| 2093 |
+
"grad_norm": 5.603676794638893,
|
| 2094 |
+
"learning_rate": 3.0239435998430374e-10,
|
| 2095 |
+
"logits/chosen": -2.0554075241088867,
|
| 2096 |
+
"logits/rejected": -1.8917900323867798,
|
| 2097 |
+
"logps/chosen": -446.63922119140625,
|
| 2098 |
+
"logps/rejected": -648.955078125,
|
| 2099 |
+
"loss": 0.2974,
|
| 2100 |
+
"rewards/accuracies": 0.887499988079071,
|
| 2101 |
+
"rewards/chosen": -1.6835094690322876,
|
| 2102 |
+
"rewards/margins": 2.1592822074890137,
|
| 2103 |
+
"rewards/margins_max": 4.456897735595703,
|
| 2104 |
+
"rewards/margins_min": 0.3158484995365143,
|
| 2105 |
+
"rewards/margins_std": 1.9191557168960571,
|
| 2106 |
+
"rewards/rejected": -3.842791795730591,
|
| 2107 |
+
"step": 1050
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 2.99,
|
| 2111 |
+
"grad_norm": 5.071722037499608,
|
| 2112 |
+
"learning_rate": 3.360539611582669e-11,
|
| 2113 |
+
"logits/chosen": -1.9219595193862915,
|
| 2114 |
+
"logits/rejected": -1.8536704778671265,
|
| 2115 |
+
"logps/chosen": -401.3136291503906,
|
| 2116 |
+
"logps/rejected": -684.7721557617188,
|
| 2117 |
+
"loss": 0.2643,
|
| 2118 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2119 |
+
"rewards/chosen": -1.4818236827850342,
|
| 2120 |
+
"rewards/margins": 2.484513998031616,
|
| 2121 |
+
"rewards/margins_max": 5.372402667999268,
|
| 2122 |
+
"rewards/margins_min": 0.5392643213272095,
|
| 2123 |
+
"rewards/margins_std": 2.2041122913360596,
|
| 2124 |
+
"rewards/rejected": -3.9663376808166504,
|
| 2125 |
+
"step": 1060
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 3.0,
|
| 2129 |
+
"step": 1065,
|
| 2130 |
+
"total_flos": 0.0,
|
| 2131 |
+
"train_loss": 0.4186206150502666,
|
| 2132 |
+
"train_runtime": 20980.8825,
|
| 2133 |
+
"train_samples_per_second": 0.812,
|
| 2134 |
+
"train_steps_per_second": 0.051
|
| 2135 |
+
}
|
| 2136 |
+
],
|
| 2137 |
+
"logging_steps": 10,
|
| 2138 |
+
"max_steps": 1065,
|
| 2139 |
+
"num_input_tokens_seen": 0,
|
| 2140 |
+
"num_train_epochs": 3,
|
| 2141 |
+
"save_steps": 100,
|
| 2142 |
+
"total_flos": 0.0,
|
| 2143 |
+
"train_batch_size": 4,
|
| 2144 |
+
"trial_name": null,
|
| 2145 |
+
"trial_params": null
|
| 2146 |
+
}
|