Model save
Browse files- README.md +88 -0
- adapter_model.safetensors +1 -1
- all_results.json +8 -0
- runs/Jul29_11-59-46_node13/events.out.tfevents.1722222409.node13.208681.0 +2 -2
- train_results.json +8 -0
- trainer_state.json +2380 -0
README.md
ADDED
|
@@ -0,0 +1,88 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 9 |
+
model-index:
|
| 10 |
+
- name: zephyr-dpop-qlora-gpt4-5e-6-epoch3
|
| 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 |
+
# zephyr-dpop-qlora-gpt4-5e-6-epoch3
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the None dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 14.5955
|
| 22 |
+
- Positive Losses: 143.9389
|
| 23 |
+
- Dpo Losses: 0.6869
|
| 24 |
+
- Rewards/chosen: -1.4291
|
| 25 |
+
- Rewards/rejected: -2.0251
|
| 26 |
+
- Rewards/accuracies: 0.6627
|
| 27 |
+
- Rewards/margins: 0.5959
|
| 28 |
+
- Rewards/margins Max: 2.2953
|
| 29 |
+
- Rewards/margins Min: -1.1073
|
| 30 |
+
- Rewards/margins Std: 1.5052
|
| 31 |
+
- Logps/rejected: -461.6887
|
| 32 |
+
- Logps/chosen: -428.1342
|
| 33 |
+
- Logits/rejected: -2.2738
|
| 34 |
+
- Logits/chosen: -2.3165
|
| 35 |
+
|
| 36 |
+
## Model description
|
| 37 |
+
|
| 38 |
+
More information needed
|
| 39 |
+
|
| 40 |
+
## Intended uses & limitations
|
| 41 |
+
|
| 42 |
+
More information needed
|
| 43 |
+
|
| 44 |
+
## Training and evaluation data
|
| 45 |
+
|
| 46 |
+
More information needed
|
| 47 |
+
|
| 48 |
+
## Training procedure
|
| 49 |
+
|
| 50 |
+
### Training hyperparameters
|
| 51 |
+
|
| 52 |
+
The following hyperparameters were used during training:
|
| 53 |
+
- learning_rate: 5e-06
|
| 54 |
+
- train_batch_size: 2
|
| 55 |
+
- eval_batch_size: 4
|
| 56 |
+
- seed: 42
|
| 57 |
+
- distributed_type: multi-GPU
|
| 58 |
+
- num_devices: 8
|
| 59 |
+
- total_train_batch_size: 16
|
| 60 |
+
- total_eval_batch_size: 32
|
| 61 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 62 |
+
- lr_scheduler_type: cosine
|
| 63 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 64 |
+
- num_epochs: 3
|
| 65 |
+
|
| 66 |
+
### Training results
|
| 67 |
+
|
| 68 |
+
| Training Loss | Epoch | Step | Validation Loss | Positive Losses | Dpo Losses | 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 |
|
| 69 |
+
|:-------------:|:-----:|:----:|:---------------:|:---------------:|:----------:|:--------------:|:----------------:|:------------------:|:---------------:|:-------------------:|:-------------------:|:-------------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 70 |
+
| 0.5432 | 0.28 | 100 | 1.5490 | 8.3683 | 0.6723 | -0.0507 | -0.1015 | 0.5992 | 0.0508 | 0.2567 | -0.1414 | 0.1757 | -269.3354 | -290.2917 | -2.6677 | -2.7099 |
|
| 71 |
+
| 0.4843 | 0.56 | 200 | 3.6354 | 28.9322 | 0.6415 | -0.2537 | -0.4297 | 0.6349 | 0.1759 | 0.7364 | -0.3533 | 0.4858 | -302.1486 | -310.5943 | -2.5589 | -2.6000 |
|
| 72 |
+
| 0.2828 | 0.85 | 300 | 6.8046 | 61.7689 | 0.6346 | -0.6003 | -0.8503 | 0.6508 | 0.2500 | 1.0085 | -0.4868 | 0.6679 | -344.2117 | -345.2526 | -2.5349 | -2.5759 |
|
| 73 |
+
| 0.3355 | 1.13 | 400 | 11.4158 | 108.7399 | 0.6572 | -1.0761 | -1.4209 | 0.6548 | 0.3447 | 1.4626 | -0.7661 | 0.9968 | -401.2702 | -392.8341 | -2.3773 | -2.4155 |
|
| 74 |
+
| 0.3438 | 1.41 | 500 | 10.6413 | 101.3525 | 0.6381 | -1.0007 | -1.3406 | 0.6865 | 0.3399 | 1.3353 | -0.6338 | 0.8805 | -393.2457 | -385.2938 | -2.4471 | -2.4907 |
|
| 75 |
+
| 0.2144 | 1.69 | 600 | 8.5896 | 79.7998 | 0.6267 | -0.7817 | -1.2135 | 0.6865 | 0.4318 | 1.5951 | -0.6661 | 1.0047 | -380.5318 | -363.3914 | -2.3029 | -2.3438 |
|
| 76 |
+
| 0.3314 | 1.97 | 700 | 11.1651 | 107.2969 | 0.6525 | -1.0595 | -1.5150 | 0.6627 | 0.4555 | 1.7776 | -0.8450 | 1.1660 | -410.6869 | -391.1705 | -2.3025 | -2.3432 |
|
| 77 |
+
| 0.1352 | 2.25 | 800 | 13.3571 | 130.9070 | 0.6700 | -1.2986 | -1.8184 | 0.6627 | 0.5198 | 2.0225 | -0.9603 | 1.3296 | -441.0237 | -415.0786 | -2.2901 | -2.3320 |
|
| 78 |
+
| 0.2348 | 2.54 | 900 | 14.7241 | 145.9081 | 0.6904 | -1.4488 | -2.0053 | 0.6706 | 0.5564 | 2.1801 | -1.0958 | 1.4586 | -459.7108 | -430.1044 | -2.2661 | -2.3085 |
|
| 79 |
+
| 0.1369 | 2.82 | 1000 | 14.5955 | 143.9389 | 0.6869 | -1.4291 | -2.0251 | 0.6627 | 0.5959 | 2.2953 | -1.1073 | 1.5052 | -461.6887 | -428.1342 | -2.2738 | -2.3165 |
|
| 80 |
+
|
| 81 |
+
|
| 82 |
+
### Framework versions
|
| 83 |
+
|
| 84 |
+
- PEFT 0.7.1
|
| 85 |
+
- Transformers 4.39.0.dev0
|
| 86 |
+
- Pytorch 2.1.2+cu121
|
| 87 |
+
- Datasets 2.14.6
|
| 88 |
+
- 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 sha256:8319769653c10d99f3730b985835d27d1eedcee4f092d2dbb20aa1d26dff630d
|
| 3 |
size 671150064
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"train_loss": 0.36563416943303856,
|
| 4 |
+
"train_runtime": 9271.2095,
|
| 5 |
+
"train_samples": 5678,
|
| 6 |
+
"train_samples_per_second": 1.837,
|
| 7 |
+
"train_steps_per_second": 0.115
|
| 8 |
+
}
|
runs/Jul29_11-59-46_node13/events.out.tfevents.1722222409.node13.208681.0
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 sha256:84877456a01902e1a32a6ffb2fe7d6f9d9d67f416aa44cecafb80e357465a873
|
| 3 |
+
size 122338
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"train_loss": 0.36563416943303856,
|
| 4 |
+
"train_runtime": 9271.2095,
|
| 5 |
+
"train_samples": 5678,
|
| 6 |
+
"train_samples_per_second": 1.837,
|
| 7 |
+
"train_steps_per_second": 0.115
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2380 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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 |
+
"dpo_losses": 0.6931471824645996,
|
| 13 |
+
"epoch": 0.0,
|
| 14 |
+
"grad_norm": 2.1182166269339366,
|
| 15 |
+
"learning_rate": 4.672897196261682e-08,
|
| 16 |
+
"logits/chosen": -2.8477635383605957,
|
| 17 |
+
"logits/rejected": -2.8469698429107666,
|
| 18 |
+
"logps/chosen": -522.6112670898438,
|
| 19 |
+
"logps/rejected": -359.48583984375,
|
| 20 |
+
"loss": 0.6931,
|
| 21 |
+
"positive_losses": 0.0,
|
| 22 |
+
"rewards/accuracies": 0.0,
|
| 23 |
+
"rewards/chosen": 0.0,
|
| 24 |
+
"rewards/margins": 0.0,
|
| 25 |
+
"rewards/margins_max": 0.0,
|
| 26 |
+
"rewards/margins_min": 0.0,
|
| 27 |
+
"rewards/margins_std": 0.0,
|
| 28 |
+
"rewards/rejected": 0.0,
|
| 29 |
+
"step": 1
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"dpo_losses": 0.6931488513946533,
|
| 33 |
+
"epoch": 0.03,
|
| 34 |
+
"grad_norm": 18.90606759984328,
|
| 35 |
+
"learning_rate": 4.6728971962616824e-07,
|
| 36 |
+
"logits/chosen": -2.921452522277832,
|
| 37 |
+
"logits/rejected": -2.7972779273986816,
|
| 38 |
+
"logps/chosen": -313.4413146972656,
|
| 39 |
+
"logps/rejected": -170.33502197265625,
|
| 40 |
+
"loss": 0.698,
|
| 41 |
+
"positive_losses": 0.056756019592285156,
|
| 42 |
+
"rewards/accuracies": 0.5555555820465088,
|
| 43 |
+
"rewards/chosen": 0.00029136601369827986,
|
| 44 |
+
"rewards/margins": -2.1007015220675385e-06,
|
| 45 |
+
"rewards/margins_max": 0.0011362881632521749,
|
| 46 |
+
"rewards/margins_min": -0.001140489592216909,
|
| 47 |
+
"rewards/margins_std": 0.0016099249478429556,
|
| 48 |
+
"rewards/rejected": 0.0002934667863883078,
|
| 49 |
+
"step": 10
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"dpo_losses": 0.6920409202575684,
|
| 53 |
+
"epoch": 0.06,
|
| 54 |
+
"grad_norm": 2.048573611790903,
|
| 55 |
+
"learning_rate": 9.345794392523365e-07,
|
| 56 |
+
"logits/chosen": -2.7631053924560547,
|
| 57 |
+
"logits/rejected": -2.7107467651367188,
|
| 58 |
+
"logps/chosen": -380.45953369140625,
|
| 59 |
+
"logps/rejected": -243.90365600585938,
|
| 60 |
+
"loss": 0.6928,
|
| 61 |
+
"positive_losses": 0.02124938927590847,
|
| 62 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 63 |
+
"rewards/chosen": 0.006503595970571041,
|
| 64 |
+
"rewards/margins": 0.0022183754481375217,
|
| 65 |
+
"rewards/margins_max": 0.004751545377075672,
|
| 66 |
+
"rewards/margins_min": -0.00031479448080062866,
|
| 67 |
+
"rewards/margins_std": 0.0035824428778141737,
|
| 68 |
+
"rewards/rejected": 0.004285220988094807,
|
| 69 |
+
"step": 20
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"dpo_losses": 0.6899846792221069,
|
| 73 |
+
"epoch": 0.08,
|
| 74 |
+
"grad_norm": 2.2814863944700297,
|
| 75 |
+
"learning_rate": 1.4018691588785047e-06,
|
| 76 |
+
"logits/chosen": -2.87530779838562,
|
| 77 |
+
"logits/rejected": -2.823071002960205,
|
| 78 |
+
"logps/chosen": -374.14593505859375,
|
| 79 |
+
"logps/rejected": -251.3402099609375,
|
| 80 |
+
"loss": 0.6884,
|
| 81 |
+
"positive_losses": 0.0,
|
| 82 |
+
"rewards/accuracies": 0.699999988079071,
|
| 83 |
+
"rewards/chosen": 0.01957734487950802,
|
| 84 |
+
"rewards/margins": 0.0063508180901408195,
|
| 85 |
+
"rewards/margins_max": 0.011502384208142757,
|
| 86 |
+
"rewards/margins_min": 0.0011992522049695253,
|
| 87 |
+
"rewards/margins_std": 0.007285414729267359,
|
| 88 |
+
"rewards/rejected": 0.013226528652012348,
|
| 89 |
+
"step": 30
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"dpo_losses": 0.6816079616546631,
|
| 93 |
+
"epoch": 0.11,
|
| 94 |
+
"grad_norm": 1.9338584641637362,
|
| 95 |
+
"learning_rate": 1.869158878504673e-06,
|
| 96 |
+
"logits/chosen": -2.732996940612793,
|
| 97 |
+
"logits/rejected": -2.7668612003326416,
|
| 98 |
+
"logps/chosen": -302.9008483886719,
|
| 99 |
+
"logps/rejected": -315.843505859375,
|
| 100 |
+
"loss": 0.6807,
|
| 101 |
+
"positive_losses": 0.0,
|
| 102 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 103 |
+
"rewards/chosen": 0.04024823382496834,
|
| 104 |
+
"rewards/margins": 0.023371612653136253,
|
| 105 |
+
"rewards/margins_max": 0.03446139022707939,
|
| 106 |
+
"rewards/margins_min": 0.01228183414787054,
|
| 107 |
+
"rewards/margins_std": 0.015683313831686974,
|
| 108 |
+
"rewards/rejected": 0.016876617446541786,
|
| 109 |
+
"step": 40
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"dpo_losses": 0.6704033613204956,
|
| 113 |
+
"epoch": 0.14,
|
| 114 |
+
"grad_norm": 2.241837849751868,
|
| 115 |
+
"learning_rate": 2.3364485981308413e-06,
|
| 116 |
+
"logits/chosen": -2.7844934463500977,
|
| 117 |
+
"logits/rejected": -2.696681022644043,
|
| 118 |
+
"logps/chosen": -238.6322784423828,
|
| 119 |
+
"logps/rejected": -172.29949951171875,
|
| 120 |
+
"loss": 0.6657,
|
| 121 |
+
"positive_losses": 0.0,
|
| 122 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 123 |
+
"rewards/chosen": 0.06590737402439117,
|
| 124 |
+
"rewards/margins": 0.046577345579862595,
|
| 125 |
+
"rewards/margins_max": 0.06979880481958389,
|
| 126 |
+
"rewards/margins_min": 0.023355895653367043,
|
| 127 |
+
"rewards/margins_std": 0.03284009173512459,
|
| 128 |
+
"rewards/rejected": 0.019330020993947983,
|
| 129 |
+
"step": 50
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"dpo_losses": 0.6440416574478149,
|
| 133 |
+
"epoch": 0.17,
|
| 134 |
+
"grad_norm": 2.210137428049826,
|
| 135 |
+
"learning_rate": 2.8037383177570094e-06,
|
| 136 |
+
"logits/chosen": -2.7359554767608643,
|
| 137 |
+
"logits/rejected": -2.6721925735473633,
|
| 138 |
+
"logps/chosen": -255.8380584716797,
|
| 139 |
+
"logps/rejected": -234.24276733398438,
|
| 140 |
+
"loss": 0.6442,
|
| 141 |
+
"positive_losses": 0.0,
|
| 142 |
+
"rewards/accuracies": 0.949999988079071,
|
| 143 |
+
"rewards/chosen": 0.11597372591495514,
|
| 144 |
+
"rewards/margins": 0.10343287885189056,
|
| 145 |
+
"rewards/margins_max": 0.15465359389781952,
|
| 146 |
+
"rewards/margins_min": 0.0522121861577034,
|
| 147 |
+
"rewards/margins_std": 0.07243702560663223,
|
| 148 |
+
"rewards/rejected": 0.012540824711322784,
|
| 149 |
+
"step": 60
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"dpo_losses": 0.6189829111099243,
|
| 153 |
+
"epoch": 0.2,
|
| 154 |
+
"grad_norm": 1.721316554071705,
|
| 155 |
+
"learning_rate": 3.2710280373831774e-06,
|
| 156 |
+
"logits/chosen": -2.6506264209747314,
|
| 157 |
+
"logits/rejected": -2.65425443649292,
|
| 158 |
+
"logps/chosen": -316.86090087890625,
|
| 159 |
+
"logps/rejected": -213.46066284179688,
|
| 160 |
+
"loss": 0.6213,
|
| 161 |
+
"positive_losses": 0.0,
|
| 162 |
+
"rewards/accuracies": 1.0,
|
| 163 |
+
"rewards/chosen": 0.17079836130142212,
|
| 164 |
+
"rewards/margins": 0.15738530457019806,
|
| 165 |
+
"rewards/margins_max": 0.23612920939922333,
|
| 166 |
+
"rewards/margins_min": 0.078641377389431,
|
| 167 |
+
"rewards/margins_std": 0.11136071383953094,
|
| 168 |
+
"rewards/rejected": 0.013413062319159508,
|
| 169 |
+
"step": 70
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"dpo_losses": 0.59584641456604,
|
| 173 |
+
"epoch": 0.23,
|
| 174 |
+
"grad_norm": 5.833494928937326,
|
| 175 |
+
"learning_rate": 3.738317757009346e-06,
|
| 176 |
+
"logits/chosen": -2.8657782077789307,
|
| 177 |
+
"logits/rejected": -2.7893776893615723,
|
| 178 |
+
"logps/chosen": -318.1582946777344,
|
| 179 |
+
"logps/rejected": -289.5111999511719,
|
| 180 |
+
"loss": 0.6052,
|
| 181 |
+
"positive_losses": 0.0,
|
| 182 |
+
"rewards/accuracies": 1.0,
|
| 183 |
+
"rewards/chosen": 0.18891094624996185,
|
| 184 |
+
"rewards/margins": 0.2127668410539627,
|
| 185 |
+
"rewards/margins_max": 0.30381911993026733,
|
| 186 |
+
"rewards/margins_min": 0.12171456962823868,
|
| 187 |
+
"rewards/margins_std": 0.12876734137535095,
|
| 188 |
+
"rewards/rejected": -0.023855898529291153,
|
| 189 |
+
"step": 80
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"dpo_losses": 0.5957463383674622,
|
| 193 |
+
"epoch": 0.25,
|
| 194 |
+
"grad_norm": 2.2476761791524478,
|
| 195 |
+
"learning_rate": 4.205607476635514e-06,
|
| 196 |
+
"logits/chosen": -2.668989658355713,
|
| 197 |
+
"logits/rejected": -2.6688218116760254,
|
| 198 |
+
"logps/chosen": -256.0438537597656,
|
| 199 |
+
"logps/rejected": -203.91128540039062,
|
| 200 |
+
"loss": 0.5735,
|
| 201 |
+
"positive_losses": 0.0,
|
| 202 |
+
"rewards/accuracies": 0.949999988079071,
|
| 203 |
+
"rewards/chosen": 0.20931395888328552,
|
| 204 |
+
"rewards/margins": 0.21628114581108093,
|
| 205 |
+
"rewards/margins_max": 0.35886088013648987,
|
| 206 |
+
"rewards/margins_min": 0.07370143383741379,
|
| 207 |
+
"rewards/margins_std": 0.20163817703723907,
|
| 208 |
+
"rewards/rejected": -0.00696719903498888,
|
| 209 |
+
"step": 90
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"dpo_losses": 0.5217684507369995,
|
| 213 |
+
"epoch": 0.28,
|
| 214 |
+
"grad_norm": 1.9962825797807737,
|
| 215 |
+
"learning_rate": 4.6728971962616825e-06,
|
| 216 |
+
"logits/chosen": -2.8356316089630127,
|
| 217 |
+
"logits/rejected": -2.7704596519470215,
|
| 218 |
+
"logps/chosen": -424.61712646484375,
|
| 219 |
+
"logps/rejected": -347.96246337890625,
|
| 220 |
+
"loss": 0.5432,
|
| 221 |
+
"positive_losses": 0.040345001965761185,
|
| 222 |
+
"rewards/accuracies": 1.0,
|
| 223 |
+
"rewards/chosen": 0.336413711309433,
|
| 224 |
+
"rewards/margins": 0.3945561945438385,
|
| 225 |
+
"rewards/margins_max": 0.5599964261054993,
|
| 226 |
+
"rewards/margins_min": 0.22911591827869415,
|
| 227 |
+
"rewards/margins_std": 0.2339678257703781,
|
| 228 |
+
"rewards/rejected": -0.058142442256212234,
|
| 229 |
+
"step": 100
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 0.28,
|
| 233 |
+
"eval_dpo_losses": 0.6723248958587646,
|
| 234 |
+
"eval_logits/chosen": -2.7099392414093018,
|
| 235 |
+
"eval_logits/rejected": -2.6677489280700684,
|
| 236 |
+
"eval_logps/chosen": -290.291748046875,
|
| 237 |
+
"eval_logps/rejected": -269.3353576660156,
|
| 238 |
+
"eval_loss": 1.5490143299102783,
|
| 239 |
+
"eval_positive_losses": 8.368339538574219,
|
| 240 |
+
"eval_rewards/accuracies": 0.5992063283920288,
|
| 241 |
+
"eval_rewards/chosen": -0.05070570856332779,
|
| 242 |
+
"eval_rewards/margins": 0.05082136392593384,
|
| 243 |
+
"eval_rewards/margins_max": 0.25667881965637207,
|
| 244 |
+
"eval_rewards/margins_min": -0.14138472080230713,
|
| 245 |
+
"eval_rewards/margins_std": 0.17566871643066406,
|
| 246 |
+
"eval_rewards/rejected": -0.10152707248926163,
|
| 247 |
+
"eval_runtime": 282.2019,
|
| 248 |
+
"eval_samples_per_second": 7.087,
|
| 249 |
+
"eval_steps_per_second": 0.223,
|
| 250 |
+
"step": 100
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"dpo_losses": 0.5300568342208862,
|
| 254 |
+
"epoch": 0.31,
|
| 255 |
+
"grad_norm": 2.4142164656542064,
|
| 256 |
+
"learning_rate": 4.999879018839288e-06,
|
| 257 |
+
"logits/chosen": -2.694153070449829,
|
| 258 |
+
"logits/rejected": -2.5792508125305176,
|
| 259 |
+
"logps/chosen": -301.54388427734375,
|
| 260 |
+
"logps/rejected": -258.5564270019531,
|
| 261 |
+
"loss": 0.5372,
|
| 262 |
+
"positive_losses": 0.238017275929451,
|
| 263 |
+
"rewards/accuracies": 1.0,
|
| 264 |
+
"rewards/chosen": 0.26155275106430054,
|
| 265 |
+
"rewards/margins": 0.3829374313354492,
|
| 266 |
+
"rewards/margins_max": 0.5613982677459717,
|
| 267 |
+
"rewards/margins_min": 0.20447655022144318,
|
| 268 |
+
"rewards/margins_std": 0.2523817718029022,
|
| 269 |
+
"rewards/rejected": -0.12138471752405167,
|
| 270 |
+
"step": 110
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"dpo_losses": 0.488577663898468,
|
| 274 |
+
"epoch": 0.34,
|
| 275 |
+
"grad_norm": 9.3268800860026,
|
| 276 |
+
"learning_rate": 4.99772856836941e-06,
|
| 277 |
+
"logits/chosen": -2.8129353523254395,
|
| 278 |
+
"logits/rejected": -2.79107403755188,
|
| 279 |
+
"logps/chosen": -337.3607177734375,
|
| 280 |
+
"logps/rejected": -347.9256286621094,
|
| 281 |
+
"loss": 0.5066,
|
| 282 |
+
"positive_losses": 0.0004665374872274697,
|
| 283 |
+
"rewards/accuracies": 1.0,
|
| 284 |
+
"rewards/chosen": 0.345800518989563,
|
| 285 |
+
"rewards/margins": 0.5089500546455383,
|
| 286 |
+
"rewards/margins_max": 0.725378692150116,
|
| 287 |
+
"rewards/margins_min": 0.29252126812934875,
|
| 288 |
+
"rewards/margins_std": 0.3060764968395233,
|
| 289 |
+
"rewards/rejected": -0.16314946115016937,
|
| 290 |
+
"step": 120
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"dpo_losses": 0.42630523443222046,
|
| 294 |
+
"epoch": 0.37,
|
| 295 |
+
"grad_norm": 6.838233431406268,
|
| 296 |
+
"learning_rate": 4.992892309373227e-06,
|
| 297 |
+
"logits/chosen": -2.64375376701355,
|
| 298 |
+
"logits/rejected": -2.5765233039855957,
|
| 299 |
+
"logps/chosen": -352.48126220703125,
|
| 300 |
+
"logps/rejected": -287.73388671875,
|
| 301 |
+
"loss": 0.4592,
|
| 302 |
+
"positive_losses": 0.0,
|
| 303 |
+
"rewards/accuracies": 1.0,
|
| 304 |
+
"rewards/chosen": 0.3849102854728699,
|
| 305 |
+
"rewards/margins": 0.679017186164856,
|
| 306 |
+
"rewards/margins_max": 0.8494507670402527,
|
| 307 |
+
"rewards/margins_min": 0.5085835456848145,
|
| 308 |
+
"rewards/margins_std": 0.24102959036827087,
|
| 309 |
+
"rewards/rejected": -0.2941069006919861,
|
| 310 |
+
"step": 130
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"dpo_losses": 0.4316105842590332,
|
| 314 |
+
"epoch": 0.39,
|
| 315 |
+
"grad_norm": 24.291983530934388,
|
| 316 |
+
"learning_rate": 4.985375442281969e-06,
|
| 317 |
+
"logits/chosen": -2.6181836128234863,
|
| 318 |
+
"logits/rejected": -2.5858638286590576,
|
| 319 |
+
"logps/chosen": -320.45770263671875,
|
| 320 |
+
"logps/rejected": -268.80950927734375,
|
| 321 |
+
"loss": 0.4791,
|
| 322 |
+
"positive_losses": 0.11783752590417862,
|
| 323 |
+
"rewards/accuracies": 0.949999988079071,
|
| 324 |
+
"rewards/chosen": 0.3075354993343353,
|
| 325 |
+
"rewards/margins": 0.6785004138946533,
|
| 326 |
+
"rewards/margins_max": 0.9275323748588562,
|
| 327 |
+
"rewards/margins_min": 0.42946839332580566,
|
| 328 |
+
"rewards/margins_std": 0.35218438506126404,
|
| 329 |
+
"rewards/rejected": -0.37096482515335083,
|
| 330 |
+
"step": 140
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"dpo_losses": 0.43038830161094666,
|
| 334 |
+
"epoch": 0.42,
|
| 335 |
+
"grad_norm": 16.31539388542976,
|
| 336 |
+
"learning_rate": 4.9751860499858175e-06,
|
| 337 |
+
"logits/chosen": -2.6524252891540527,
|
| 338 |
+
"logits/rejected": -2.582724094390869,
|
| 339 |
+
"logps/chosen": -269.40289306640625,
|
| 340 |
+
"logps/rejected": -286.25921630859375,
|
| 341 |
+
"loss": 0.4405,
|
| 342 |
+
"positive_losses": 0.45142823457717896,
|
| 343 |
+
"rewards/accuracies": 1.0,
|
| 344 |
+
"rewards/chosen": 0.27419334650039673,
|
| 345 |
+
"rewards/margins": 0.6590893268585205,
|
| 346 |
+
"rewards/margins_max": 0.8764106631278992,
|
| 347 |
+
"rewards/margins_min": 0.44176802039146423,
|
| 348 |
+
"rewards/margins_std": 0.307338684797287,
|
| 349 |
+
"rewards/rejected": -0.3848959803581238,
|
| 350 |
+
"step": 150
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"dpo_losses": 0.40865588188171387,
|
| 354 |
+
"epoch": 0.45,
|
| 355 |
+
"grad_norm": 9.196914105554178,
|
| 356 |
+
"learning_rate": 4.962335089142376e-06,
|
| 357 |
+
"logits/chosen": -2.671811580657959,
|
| 358 |
+
"logits/rejected": -2.557887554168701,
|
| 359 |
+
"logps/chosen": -298.1056823730469,
|
| 360 |
+
"logps/rejected": -286.0025329589844,
|
| 361 |
+
"loss": 0.4273,
|
| 362 |
+
"positive_losses": 1.3531090021133423,
|
| 363 |
+
"rewards/accuracies": 1.0,
|
| 364 |
+
"rewards/chosen": 0.3361397683620453,
|
| 365 |
+
"rewards/margins": 0.7648395299911499,
|
| 366 |
+
"rewards/margins_max": 1.1010096073150635,
|
| 367 |
+
"rewards/margins_min": 0.42866945266723633,
|
| 368 |
+
"rewards/margins_std": 0.47541624307632446,
|
| 369 |
+
"rewards/rejected": -0.4286997318267822,
|
| 370 |
+
"step": 160
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"dpo_losses": 0.36792081594467163,
|
| 374 |
+
"epoch": 0.48,
|
| 375 |
+
"grad_norm": 3.4842547066626572,
|
| 376 |
+
"learning_rate": 4.946836378394967e-06,
|
| 377 |
+
"logits/chosen": -2.719280242919922,
|
| 378 |
+
"logits/rejected": -2.560034990310669,
|
| 379 |
+
"logps/chosen": -358.5220642089844,
|
| 380 |
+
"logps/rejected": -287.3617858886719,
|
| 381 |
+
"loss": 0.4612,
|
| 382 |
+
"positive_losses": 0.0,
|
| 383 |
+
"rewards/accuracies": 1.0,
|
| 384 |
+
"rewards/chosen": 0.4195915162563324,
|
| 385 |
+
"rewards/margins": 0.9577536582946777,
|
| 386 |
+
"rewards/margins_max": 1.3000526428222656,
|
| 387 |
+
"rewards/margins_min": 0.6154545545578003,
|
| 388 |
+
"rewards/margins_std": 0.48408395051956177,
|
| 389 |
+
"rewards/rejected": -0.5381620526313782,
|
| 390 |
+
"step": 170
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"dpo_losses": 0.3783532977104187,
|
| 394 |
+
"epoch": 0.51,
|
| 395 |
+
"grad_norm": 2.2304162655616295,
|
| 396 |
+
"learning_rate": 4.928706583513441e-06,
|
| 397 |
+
"logits/chosen": -2.599997043609619,
|
| 398 |
+
"logits/rejected": -2.5286357402801514,
|
| 399 |
+
"logps/chosen": -297.5575866699219,
|
| 400 |
+
"logps/rejected": -439.0782165527344,
|
| 401 |
+
"loss": 0.4178,
|
| 402 |
+
"positive_losses": 0.7186470031738281,
|
| 403 |
+
"rewards/accuracies": 1.0,
|
| 404 |
+
"rewards/chosen": 0.26578420400619507,
|
| 405 |
+
"rewards/margins": 0.8594935536384583,
|
| 406 |
+
"rewards/margins_max": 1.1510531902313232,
|
| 407 |
+
"rewards/margins_min": 0.5679339170455933,
|
| 408 |
+
"rewards/margins_std": 0.4123275876045227,
|
| 409 |
+
"rewards/rejected": -0.5937093496322632,
|
| 410 |
+
"step": 180
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"dpo_losses": 0.3120216727256775,
|
| 414 |
+
"epoch": 0.54,
|
| 415 |
+
"grad_norm": 2.502842717440193,
|
| 416 |
+
"learning_rate": 4.907965199473471e-06,
|
| 417 |
+
"logits/chosen": -2.5457229614257812,
|
| 418 |
+
"logits/rejected": -2.3784427642822266,
|
| 419 |
+
"logps/chosen": -403.69097900390625,
|
| 420 |
+
"logps/rejected": -277.13079833984375,
|
| 421 |
+
"loss": 0.4291,
|
| 422 |
+
"positive_losses": 0.1399887055158615,
|
| 423 |
+
"rewards/accuracies": 1.0,
|
| 424 |
+
"rewards/chosen": 0.524212658405304,
|
| 425 |
+
"rewards/margins": 1.0702520608901978,
|
| 426 |
+
"rewards/margins_max": 1.2478736639022827,
|
| 427 |
+
"rewards/margins_min": 0.892630398273468,
|
| 428 |
+
"rewards/margins_std": 0.25119495391845703,
|
| 429 |
+
"rewards/rejected": -0.546039342880249,
|
| 430 |
+
"step": 190
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"dpo_losses": 0.340221643447876,
|
| 434 |
+
"epoch": 0.56,
|
| 435 |
+
"grad_norm": 39.20326200653621,
|
| 436 |
+
"learning_rate": 4.884634529493591e-06,
|
| 437 |
+
"logits/chosen": -2.686732769012451,
|
| 438 |
+
"logits/rejected": -2.5898447036743164,
|
| 439 |
+
"logps/chosen": -290.22259521484375,
|
| 440 |
+
"logps/rejected": -267.71923828125,
|
| 441 |
+
"loss": 0.4843,
|
| 442 |
+
"positive_losses": 1.2471590042114258,
|
| 443 |
+
"rewards/accuracies": 0.949999988079071,
|
| 444 |
+
"rewards/chosen": 0.405730664730072,
|
| 445 |
+
"rewards/margins": 1.0374891757965088,
|
| 446 |
+
"rewards/margins_max": 1.3196719884872437,
|
| 447 |
+
"rewards/margins_min": 0.7553063035011292,
|
| 448 |
+
"rewards/margins_std": 0.3990669250488281,
|
| 449 |
+
"rewards/rejected": -0.6317585110664368,
|
| 450 |
+
"step": 200
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.56,
|
| 454 |
+
"eval_dpo_losses": 0.6414885520935059,
|
| 455 |
+
"eval_logits/chosen": -2.6000208854675293,
|
| 456 |
+
"eval_logits/rejected": -2.5589327812194824,
|
| 457 |
+
"eval_logps/chosen": -310.5942687988281,
|
| 458 |
+
"eval_logps/rejected": -302.1485595703125,
|
| 459 |
+
"eval_loss": 3.6353535652160645,
|
| 460 |
+
"eval_positive_losses": 28.93216323852539,
|
| 461 |
+
"eval_rewards/accuracies": 0.6349206566810608,
|
| 462 |
+
"eval_rewards/chosen": -0.2537309527397156,
|
| 463 |
+
"eval_rewards/margins": 0.17592783272266388,
|
| 464 |
+
"eval_rewards/margins_max": 0.7364258766174316,
|
| 465 |
+
"eval_rewards/margins_min": -0.35331711173057556,
|
| 466 |
+
"eval_rewards/margins_std": 0.48577553033828735,
|
| 467 |
+
"eval_rewards/rejected": -0.42965877056121826,
|
| 468 |
+
"eval_runtime": 280.7482,
|
| 469 |
+
"eval_samples_per_second": 7.124,
|
| 470 |
+
"eval_steps_per_second": 0.224,
|
| 471 |
+
"step": 200
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"dpo_losses": 0.4035143256187439,
|
| 475 |
+
"epoch": 0.59,
|
| 476 |
+
"grad_norm": 2.0960462134704727,
|
| 477 |
+
"learning_rate": 4.858739661052539e-06,
|
| 478 |
+
"logits/chosen": -2.449962854385376,
|
| 479 |
+
"logits/rejected": -2.364846706390381,
|
| 480 |
+
"logps/chosen": -316.26300048828125,
|
| 481 |
+
"logps/rejected": -311.43115234375,
|
| 482 |
+
"loss": 0.3993,
|
| 483 |
+
"positive_losses": 0.28444308042526245,
|
| 484 |
+
"rewards/accuracies": 0.949999988079071,
|
| 485 |
+
"rewards/chosen": 0.4057347774505615,
|
| 486 |
+
"rewards/margins": 0.8319534063339233,
|
| 487 |
+
"rewards/margins_max": 1.267531156539917,
|
| 488 |
+
"rewards/margins_min": 0.3963755965232849,
|
| 489 |
+
"rewards/margins_std": 0.6160000562667847,
|
| 490 |
+
"rewards/rejected": -0.4262186586856842,
|
| 491 |
+
"step": 210
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"dpo_losses": 0.2952147424221039,
|
| 495 |
+
"epoch": 0.62,
|
| 496 |
+
"grad_norm": 2.161286043094922,
|
| 497 |
+
"learning_rate": 4.830308438912687e-06,
|
| 498 |
+
"logits/chosen": -2.7107901573181152,
|
| 499 |
+
"logits/rejected": -2.560708522796631,
|
| 500 |
+
"logps/chosen": -364.57659912109375,
|
| 501 |
+
"logps/rejected": -361.52545166015625,
|
| 502 |
+
"loss": 0.3728,
|
| 503 |
+
"positive_losses": 0.0,
|
| 504 |
+
"rewards/accuracies": 1.0,
|
| 505 |
+
"rewards/chosen": 0.4469282031059265,
|
| 506 |
+
"rewards/margins": 1.2792309522628784,
|
| 507 |
+
"rewards/margins_max": 1.8157857656478882,
|
| 508 |
+
"rewards/margins_min": 0.7426761388778687,
|
| 509 |
+
"rewards/margins_std": 0.7588031888008118,
|
| 510 |
+
"rewards/rejected": -0.8323026895523071,
|
| 511 |
+
"step": 220
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"dpo_losses": 0.26847127079963684,
|
| 515 |
+
"epoch": 0.65,
|
| 516 |
+
"grad_norm": 14.213158348490241,
|
| 517 |
+
"learning_rate": 4.799371435178544e-06,
|
| 518 |
+
"logits/chosen": -2.694364070892334,
|
| 519 |
+
"logits/rejected": -2.611956834793091,
|
| 520 |
+
"logps/chosen": -353.65203857421875,
|
| 521 |
+
"logps/rejected": -447.9458923339844,
|
| 522 |
+
"loss": 0.4538,
|
| 523 |
+
"positive_losses": 0.512774646282196,
|
| 524 |
+
"rewards/accuracies": 1.0,
|
| 525 |
+
"rewards/chosen": 0.2710634768009186,
|
| 526 |
+
"rewards/margins": 1.3846604824066162,
|
| 527 |
+
"rewards/margins_max": 1.9593979120254517,
|
| 528 |
+
"rewards/margins_min": 0.809922993183136,
|
| 529 |
+
"rewards/margins_std": 0.8128012418746948,
|
| 530 |
+
"rewards/rejected": -1.113596796989441,
|
| 531 |
+
"step": 230
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"dpo_losses": 0.3169272541999817,
|
| 535 |
+
"epoch": 0.68,
|
| 536 |
+
"grad_norm": 19.218706332521528,
|
| 537 |
+
"learning_rate": 4.765961916422575e-06,
|
| 538 |
+
"logits/chosen": -2.664609670639038,
|
| 539 |
+
"logits/rejected": -2.5199811458587646,
|
| 540 |
+
"logps/chosen": -309.8558349609375,
|
| 541 |
+
"logps/rejected": -378.1949157714844,
|
| 542 |
+
"loss": 0.4789,
|
| 543 |
+
"positive_losses": 2.6823112964630127,
|
| 544 |
+
"rewards/accuracies": 0.949999988079071,
|
| 545 |
+
"rewards/chosen": 0.3129195272922516,
|
| 546 |
+
"rewards/margins": 1.156752586364746,
|
| 547 |
+
"rewards/margins_max": 1.4397625923156738,
|
| 548 |
+
"rewards/margins_min": 0.8737425804138184,
|
| 549 |
+
"rewards/margins_std": 0.4002366065979004,
|
| 550 |
+
"rewards/rejected": -0.8438330888748169,
|
| 551 |
+
"step": 240
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"dpo_losses": 0.2593916356563568,
|
| 555 |
+
"epoch": 0.7,
|
| 556 |
+
"grad_norm": 10.943604253629541,
|
| 557 |
+
"learning_rate": 4.730115807913627e-06,
|
| 558 |
+
"logits/chosen": -2.6513824462890625,
|
| 559 |
+
"logits/rejected": -2.482832431793213,
|
| 560 |
+
"logps/chosen": -369.9505920410156,
|
| 561 |
+
"logps/rejected": -348.71759033203125,
|
| 562 |
+
"loss": 0.3659,
|
| 563 |
+
"positive_losses": 0.7899643182754517,
|
| 564 |
+
"rewards/accuracies": 1.0,
|
| 565 |
+
"rewards/chosen": 0.45276421308517456,
|
| 566 |
+
"rewards/margins": 1.3846436738967896,
|
| 567 |
+
"rewards/margins_max": 1.6470266580581665,
|
| 568 |
+
"rewards/margins_min": 1.1222608089447021,
|
| 569 |
+
"rewards/margins_std": 0.3710656762123108,
|
| 570 |
+
"rewards/rejected": -0.9318795204162598,
|
| 571 |
+
"step": 250
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"dpo_losses": 0.3340635895729065,
|
| 575 |
+
"epoch": 0.73,
|
| 576 |
+
"grad_norm": 12.721859863392186,
|
| 577 |
+
"learning_rate": 4.691871654986485e-06,
|
| 578 |
+
"logits/chosen": -2.592149019241333,
|
| 579 |
+
"logits/rejected": -2.553668975830078,
|
| 580 |
+
"logps/chosen": -290.64971923828125,
|
| 581 |
+
"logps/rejected": -312.02752685546875,
|
| 582 |
+
"loss": 0.5868,
|
| 583 |
+
"positive_losses": 1.0232231616973877,
|
| 584 |
+
"rewards/accuracies": 1.0,
|
| 585 |
+
"rewards/chosen": 0.3117257356643677,
|
| 586 |
+
"rewards/margins": 1.1895887851715088,
|
| 587 |
+
"rewards/margins_max": 1.7240755558013916,
|
| 588 |
+
"rewards/margins_min": 0.6551022529602051,
|
| 589 |
+
"rewards/margins_std": 0.755878210067749,
|
| 590 |
+
"rewards/rejected": -0.8778631091117859,
|
| 591 |
+
"step": 260
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"dpo_losses": 0.2871754467487335,
|
| 595 |
+
"epoch": 0.76,
|
| 596 |
+
"grad_norm": 20.74321957766599,
|
| 597 |
+
"learning_rate": 4.651270581594054e-06,
|
| 598 |
+
"logits/chosen": -2.6253020763397217,
|
| 599 |
+
"logits/rejected": -2.486907482147217,
|
| 600 |
+
"logps/chosen": -381.5191955566406,
|
| 601 |
+
"logps/rejected": -312.0382995605469,
|
| 602 |
+
"loss": 0.4914,
|
| 603 |
+
"positive_losses": 0.877484142780304,
|
| 604 |
+
"rewards/accuracies": 1.0,
|
| 605 |
+
"rewards/chosen": 0.4747743606567383,
|
| 606 |
+
"rewards/margins": 1.2697714567184448,
|
| 607 |
+
"rewards/margins_max": 1.6828867197036743,
|
| 608 |
+
"rewards/margins_min": 0.8566561937332153,
|
| 609 |
+
"rewards/margins_std": 0.5842332243919373,
|
| 610 |
+
"rewards/rejected": -0.7949970960617065,
|
| 611 |
+
"step": 270
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"dpo_losses": 0.325231671333313,
|
| 615 |
+
"epoch": 0.79,
|
| 616 |
+
"grad_norm": 24.63939637076414,
|
| 617 |
+
"learning_rate": 4.6083562460867545e-06,
|
| 618 |
+
"logits/chosen": -2.4967987537384033,
|
| 619 |
+
"logits/rejected": -2.4599080085754395,
|
| 620 |
+
"logps/chosen": -295.0630798339844,
|
| 621 |
+
"logps/rejected": -352.3880615234375,
|
| 622 |
+
"loss": 0.4194,
|
| 623 |
+
"positive_losses": 1.5066649913787842,
|
| 624 |
+
"rewards/accuracies": 0.949999988079071,
|
| 625 |
+
"rewards/chosen": 0.34013646841049194,
|
| 626 |
+
"rewards/margins": 1.2071036100387573,
|
| 627 |
+
"rewards/margins_max": 1.8113868236541748,
|
| 628 |
+
"rewards/margins_min": 0.6028203964233398,
|
| 629 |
+
"rewards/margins_std": 0.8545855283737183,
|
| 630 |
+
"rewards/rejected": -0.8669670820236206,
|
| 631 |
+
"step": 280
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"dpo_losses": 0.29185962677001953,
|
| 635 |
+
"epoch": 0.82,
|
| 636 |
+
"grad_norm": 49.4876607292477,
|
| 637 |
+
"learning_rate": 4.563174794266684e-06,
|
| 638 |
+
"logits/chosen": -2.725825071334839,
|
| 639 |
+
"logits/rejected": -2.5793709754943848,
|
| 640 |
+
"logps/chosen": -316.76153564453125,
|
| 641 |
+
"logps/rejected": -357.36334228515625,
|
| 642 |
+
"loss": 0.4944,
|
| 643 |
+
"positive_losses": 1.7035411596298218,
|
| 644 |
+
"rewards/accuracies": 1.0,
|
| 645 |
+
"rewards/chosen": 0.45084109902381897,
|
| 646 |
+
"rewards/margins": 1.3104431629180908,
|
| 647 |
+
"rewards/margins_max": 1.7685630321502686,
|
| 648 |
+
"rewards/margins_min": 0.8523231744766235,
|
| 649 |
+
"rewards/margins_std": 0.6478795409202576,
|
| 650 |
+
"rewards/rejected": -0.859602153301239,
|
| 651 |
+
"step": 290
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"dpo_losses": 0.24671301245689392,
|
| 655 |
+
"epoch": 0.85,
|
| 656 |
+
"grad_norm": 38.65216195794793,
|
| 657 |
+
"learning_rate": 4.5157748097670125e-06,
|
| 658 |
+
"logits/chosen": -2.6136484146118164,
|
| 659 |
+
"logits/rejected": -2.5544955730438232,
|
| 660 |
+
"logps/chosen": -294.6103820800781,
|
| 661 |
+
"logps/rejected": -422.00531005859375,
|
| 662 |
+
"loss": 0.2828,
|
| 663 |
+
"positive_losses": 0.0,
|
| 664 |
+
"rewards/accuracies": 1.0,
|
| 665 |
+
"rewards/chosen": 0.43132978677749634,
|
| 666 |
+
"rewards/margins": 1.5242297649383545,
|
| 667 |
+
"rewards/margins_max": 1.9953060150146484,
|
| 668 |
+
"rewards/margins_min": 1.0531535148620605,
|
| 669 |
+
"rewards/margins_std": 0.6662023067474365,
|
| 670 |
+
"rewards/rejected": -1.092900037765503,
|
| 671 |
+
"step": 300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 0.85,
|
| 675 |
+
"eval_dpo_losses": 0.6346198320388794,
|
| 676 |
+
"eval_logits/chosen": -2.5758955478668213,
|
| 677 |
+
"eval_logits/rejected": -2.534869909286499,
|
| 678 |
+
"eval_logps/chosen": -345.25262451171875,
|
| 679 |
+
"eval_logps/rejected": -344.2117004394531,
|
| 680 |
+
"eval_loss": 6.804558753967285,
|
| 681 |
+
"eval_positive_losses": 61.76890182495117,
|
| 682 |
+
"eval_rewards/accuracies": 0.6507936716079712,
|
| 683 |
+
"eval_rewards/chosen": -0.6003143787384033,
|
| 684 |
+
"eval_rewards/margins": 0.24997644126415253,
|
| 685 |
+
"eval_rewards/margins_max": 1.0084712505340576,
|
| 686 |
+
"eval_rewards/margins_min": -0.48679542541503906,
|
| 687 |
+
"eval_rewards/margins_std": 0.6679101586341858,
|
| 688 |
+
"eval_rewards/rejected": -0.8502907156944275,
|
| 689 |
+
"eval_runtime": 281.0888,
|
| 690 |
+
"eval_samples_per_second": 7.115,
|
| 691 |
+
"eval_steps_per_second": 0.224,
|
| 692 |
+
"step": 300
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"dpo_losses": 0.2701479196548462,
|
| 696 |
+
"epoch": 0.87,
|
| 697 |
+
"grad_norm": 2.3545752201762977,
|
| 698 |
+
"learning_rate": 4.466207261809989e-06,
|
| 699 |
+
"logits/chosen": -2.7792999744415283,
|
| 700 |
+
"logits/rejected": -2.558169364929199,
|
| 701 |
+
"logps/chosen": -376.7308044433594,
|
| 702 |
+
"logps/rejected": -358.2815246582031,
|
| 703 |
+
"loss": 0.4216,
|
| 704 |
+
"positive_losses": 0.41081467270851135,
|
| 705 |
+
"rewards/accuracies": 1.0,
|
| 706 |
+
"rewards/chosen": 0.47516530752182007,
|
| 707 |
+
"rewards/margins": 1.4481419324874878,
|
| 708 |
+
"rewards/margins_max": 1.9792436361312866,
|
| 709 |
+
"rewards/margins_min": 0.9170401692390442,
|
| 710 |
+
"rewards/margins_std": 0.7510912418365479,
|
| 711 |
+
"rewards/rejected": -0.972976565361023,
|
| 712 |
+
"step": 310
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"dpo_losses": 0.2924317717552185,
|
| 716 |
+
"epoch": 0.9,
|
| 717 |
+
"grad_norm": 12.007334692835927,
|
| 718 |
+
"learning_rate": 4.414525450399713e-06,
|
| 719 |
+
"logits/chosen": -2.6015143394470215,
|
| 720 |
+
"logits/rejected": -2.5146775245666504,
|
| 721 |
+
"logps/chosen": -315.1227111816406,
|
| 722 |
+
"logps/rejected": -350.9872131347656,
|
| 723 |
+
"loss": 0.3361,
|
| 724 |
+
"positive_losses": 2.5894155502319336,
|
| 725 |
+
"rewards/accuracies": 0.949999988079071,
|
| 726 |
+
"rewards/chosen": 0.3113010823726654,
|
| 727 |
+
"rewards/margins": 1.4402223825454712,
|
| 728 |
+
"rewards/margins_max": 2.083007335662842,
|
| 729 |
+
"rewards/margins_min": 0.7974374890327454,
|
| 730 |
+
"rewards/margins_std": 0.9090349078178406,
|
| 731 |
+
"rewards/rejected": -1.128921389579773,
|
| 732 |
+
"step": 320
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"dpo_losses": 0.2676263153553009,
|
| 736 |
+
"epoch": 0.93,
|
| 737 |
+
"grad_norm": 29.603619163509805,
|
| 738 |
+
"learning_rate": 4.360784949008615e-06,
|
| 739 |
+
"logits/chosen": -2.705920934677124,
|
| 740 |
+
"logits/rejected": -2.5720715522766113,
|
| 741 |
+
"logps/chosen": -338.70623779296875,
|
| 742 |
+
"logps/rejected": -376.7459716796875,
|
| 743 |
+
"loss": 0.4384,
|
| 744 |
+
"positive_losses": 2.4138710498809814,
|
| 745 |
+
"rewards/accuracies": 1.0,
|
| 746 |
+
"rewards/chosen": 0.40980419516563416,
|
| 747 |
+
"rewards/margins": 1.7029374837875366,
|
| 748 |
+
"rewards/margins_max": 2.3179752826690674,
|
| 749 |
+
"rewards/margins_min": 1.087899923324585,
|
| 750 |
+
"rewards/margins_std": 0.8697945475578308,
|
| 751 |
+
"rewards/rejected": -1.2931334972381592,
|
| 752 |
+
"step": 330
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"dpo_losses": 0.3142815828323364,
|
| 756 |
+
"epoch": 0.96,
|
| 757 |
+
"grad_norm": 18.502276306325797,
|
| 758 |
+
"learning_rate": 4.30504354481929e-06,
|
| 759 |
+
"logits/chosen": -2.5686209201812744,
|
| 760 |
+
"logits/rejected": -2.424360990524292,
|
| 761 |
+
"logps/chosen": -333.12530517578125,
|
| 762 |
+
"logps/rejected": -292.06707763671875,
|
| 763 |
+
"loss": 0.3621,
|
| 764 |
+
"positive_losses": 0.5738517642021179,
|
| 765 |
+
"rewards/accuracies": 1.0,
|
| 766 |
+
"rewards/chosen": 0.3537023365497589,
|
| 767 |
+
"rewards/margins": 1.2313454151153564,
|
| 768 |
+
"rewards/margins_max": 1.7322998046875,
|
| 769 |
+
"rewards/margins_min": 0.7303910255432129,
|
| 770 |
+
"rewards/margins_std": 0.7084565758705139,
|
| 771 |
+
"rewards/rejected": -0.8776431083679199,
|
| 772 |
+
"step": 340
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"dpo_losses": 0.18595710396766663,
|
| 776 |
+
"epoch": 0.99,
|
| 777 |
+
"grad_norm": 137.9970915114699,
|
| 778 |
+
"learning_rate": 4.247361176585904e-06,
|
| 779 |
+
"logits/chosen": -2.5403497219085693,
|
| 780 |
+
"logits/rejected": -2.4443917274475098,
|
| 781 |
+
"logps/chosen": -386.3079833984375,
|
| 782 |
+
"logps/rejected": -441.9620056152344,
|
| 783 |
+
"loss": 0.4456,
|
| 784 |
+
"positive_losses": 0.9461520910263062,
|
| 785 |
+
"rewards/accuracies": 1.0,
|
| 786 |
+
"rewards/chosen": 0.4209546446800232,
|
| 787 |
+
"rewards/margins": 1.8566910028457642,
|
| 788 |
+
"rewards/margins_max": 2.370370388031006,
|
| 789 |
+
"rewards/margins_min": 1.3430118560791016,
|
| 790 |
+
"rewards/margins_std": 0.7264522910118103,
|
| 791 |
+
"rewards/rejected": -1.4357364177703857,
|
| 792 |
+
"step": 350
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"dpo_losses": 0.24505066871643066,
|
| 796 |
+
"epoch": 1.01,
|
| 797 |
+
"grad_norm": 3.181861527878417,
|
| 798 |
+
"learning_rate": 4.187799870182038e-06,
|
| 799 |
+
"logits/chosen": -2.5239622592926025,
|
| 800 |
+
"logits/rejected": -2.4095091819763184,
|
| 801 |
+
"logps/chosen": -305.76092529296875,
|
| 802 |
+
"logps/rejected": -311.29388427734375,
|
| 803 |
+
"loss": 0.3141,
|
| 804 |
+
"positive_losses": 0.24010220170021057,
|
| 805 |
+
"rewards/accuracies": 1.0,
|
| 806 |
+
"rewards/chosen": 0.4044499397277832,
|
| 807 |
+
"rewards/margins": 1.4896382093429565,
|
| 808 |
+
"rewards/margins_max": 2.010220527648926,
|
| 809 |
+
"rewards/margins_min": 0.9690560102462769,
|
| 810 |
+
"rewards/margins_std": 0.7362144589424133,
|
| 811 |
+
"rewards/rejected": -1.0851882696151733,
|
| 812 |
+
"step": 360
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"dpo_losses": 0.24201233685016632,
|
| 816 |
+
"epoch": 1.04,
|
| 817 |
+
"grad_norm": 54.166595052035994,
|
| 818 |
+
"learning_rate": 4.1264236719042365e-06,
|
| 819 |
+
"logits/chosen": -2.362025737762451,
|
| 820 |
+
"logits/rejected": -2.3736584186553955,
|
| 821 |
+
"logps/chosen": -307.57098388671875,
|
| 822 |
+
"logps/rejected": -404.64447021484375,
|
| 823 |
+
"loss": 0.3927,
|
| 824 |
+
"positive_losses": 0.0,
|
| 825 |
+
"rewards/accuracies": 1.0,
|
| 826 |
+
"rewards/chosen": 0.44086480140686035,
|
| 827 |
+
"rewards/margins": 1.8494741916656494,
|
| 828 |
+
"rewards/margins_max": 2.651545763015747,
|
| 829 |
+
"rewards/margins_min": 1.0474025011062622,
|
| 830 |
+
"rewards/margins_std": 1.1343004703521729,
|
| 831 |
+
"rewards/rejected": -1.40860915184021,
|
| 832 |
+
"step": 370
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"dpo_losses": 0.22759906947612762,
|
| 836 |
+
"epoch": 1.07,
|
| 837 |
+
"grad_norm": 38.27158679057807,
|
| 838 |
+
"learning_rate": 4.063298579603001e-06,
|
| 839 |
+
"logits/chosen": -2.472991466522217,
|
| 840 |
+
"logits/rejected": -2.2599282264709473,
|
| 841 |
+
"logps/chosen": -351.54620361328125,
|
| 842 |
+
"logps/rejected": -319.0135803222656,
|
| 843 |
+
"loss": 0.3817,
|
| 844 |
+
"positive_losses": 3.08628511428833,
|
| 845 |
+
"rewards/accuracies": 1.0,
|
| 846 |
+
"rewards/chosen": 0.3943432867527008,
|
| 847 |
+
"rewards/margins": 1.7396957874298096,
|
| 848 |
+
"rewards/margins_max": 2.3970978260040283,
|
| 849 |
+
"rewards/margins_min": 1.0822933912277222,
|
| 850 |
+
"rewards/margins_std": 0.9297070503234863,
|
| 851 |
+
"rewards/rejected": -1.3453524112701416,
|
| 852 |
+
"step": 380
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"dpo_losses": 0.134817972779274,
|
| 856 |
+
"epoch": 1.1,
|
| 857 |
+
"grad_norm": 2.791305623027733,
|
| 858 |
+
"learning_rate": 3.998492471715272e-06,
|
| 859 |
+
"logits/chosen": -2.496638059616089,
|
| 860 |
+
"logits/rejected": -2.455773115158081,
|
| 861 |
+
"logps/chosen": -349.44073486328125,
|
| 862 |
+
"logps/rejected": -536.924560546875,
|
| 863 |
+
"loss": 0.3072,
|
| 864 |
+
"positive_losses": 0.0,
|
| 865 |
+
"rewards/accuracies": 1.0,
|
| 866 |
+
"rewards/chosen": 0.4807213246822357,
|
| 867 |
+
"rewards/margins": 2.533548355102539,
|
| 868 |
+
"rewards/margins_max": 3.3716654777526855,
|
| 869 |
+
"rewards/margins_min": 1.695431113243103,
|
| 870 |
+
"rewards/margins_std": 1.185276746749878,
|
| 871 |
+
"rewards/rejected": -2.0528271198272705,
|
| 872 |
+
"step": 390
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"dpo_losses": 0.23253802955150604,
|
| 876 |
+
"epoch": 1.13,
|
| 877 |
+
"grad_norm": 1.567698262162245,
|
| 878 |
+
"learning_rate": 3.932075034274723e-06,
|
| 879 |
+
"logits/chosen": -2.385937213897705,
|
| 880 |
+
"logits/rejected": -2.3955626487731934,
|
| 881 |
+
"logps/chosen": -266.9366455078125,
|
| 882 |
+
"logps/rejected": -367.39764404296875,
|
| 883 |
+
"loss": 0.3355,
|
| 884 |
+
"positive_losses": 0.33393630385398865,
|
| 885 |
+
"rewards/accuracies": 1.0,
|
| 886 |
+
"rewards/chosen": 0.3180273175239563,
|
| 887 |
+
"rewards/margins": 1.719641923904419,
|
| 888 |
+
"rewards/margins_max": 2.0190768241882324,
|
| 889 |
+
"rewards/margins_min": 1.4202073812484741,
|
| 890 |
+
"rewards/margins_std": 0.4234645366668701,
|
| 891 |
+
"rewards/rejected": -1.401614785194397,
|
| 892 |
+
"step": 400
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 1.13,
|
| 896 |
+
"eval_dpo_losses": 0.6571967601776123,
|
| 897 |
+
"eval_logits/chosen": -2.415469169616699,
|
| 898 |
+
"eval_logits/rejected": -2.3773038387298584,
|
| 899 |
+
"eval_logps/chosen": -392.8340759277344,
|
| 900 |
+
"eval_logps/rejected": -401.27020263671875,
|
| 901 |
+
"eval_loss": 11.415841102600098,
|
| 902 |
+
"eval_positive_losses": 108.73988342285156,
|
| 903 |
+
"eval_rewards/accuracies": 0.6547619104385376,
|
| 904 |
+
"eval_rewards/chosen": -1.0761287212371826,
|
| 905 |
+
"eval_rewards/margins": 0.34474655985832214,
|
| 906 |
+
"eval_rewards/margins_max": 1.462611198425293,
|
| 907 |
+
"eval_rewards/margins_min": -0.7661140561103821,
|
| 908 |
+
"eval_rewards/margins_std": 0.9968340992927551,
|
| 909 |
+
"eval_rewards/rejected": -1.4208753108978271,
|
| 910 |
+
"eval_runtime": 282.7743,
|
| 911 |
+
"eval_samples_per_second": 7.073,
|
| 912 |
+
"eval_steps_per_second": 0.223,
|
| 913 |
+
"step": 400
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"dpo_losses": 0.23369018733501434,
|
| 917 |
+
"epoch": 1.15,
|
| 918 |
+
"grad_norm": 1.6752309896133104,
|
| 919 |
+
"learning_rate": 3.864117685978339e-06,
|
| 920 |
+
"logits/chosen": -2.4605586528778076,
|
| 921 |
+
"logits/rejected": -2.4261014461517334,
|
| 922 |
+
"logps/chosen": -264.14556884765625,
|
| 923 |
+
"logps/rejected": -375.3333740234375,
|
| 924 |
+
"loss": 0.3689,
|
| 925 |
+
"positive_losses": 3.3056740760803223,
|
| 926 |
+
"rewards/accuracies": 0.949999988079071,
|
| 927 |
+
"rewards/chosen": 0.3054488003253937,
|
| 928 |
+
"rewards/margins": 1.9768092632293701,
|
| 929 |
+
"rewards/margins_max": 2.963430166244507,
|
| 930 |
+
"rewards/margins_min": 0.990188479423523,
|
| 931 |
+
"rewards/margins_std": 1.3952926397323608,
|
| 932 |
+
"rewards/rejected": -1.6713603734970093,
|
| 933 |
+
"step": 410
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"dpo_losses": 0.1611286997795105,
|
| 937 |
+
"epoch": 1.18,
|
| 938 |
+
"grad_norm": 19.99801702205629,
|
| 939 |
+
"learning_rate": 3.794693501389861e-06,
|
| 940 |
+
"logits/chosen": -2.549091100692749,
|
| 941 |
+
"logits/rejected": -2.4567952156066895,
|
| 942 |
+
"logps/chosen": -349.955810546875,
|
| 943 |
+
"logps/rejected": -440.55987548828125,
|
| 944 |
+
"loss": 0.3233,
|
| 945 |
+
"positive_losses": 0.5550443530082703,
|
| 946 |
+
"rewards/accuracies": 1.0,
|
| 947 |
+
"rewards/chosen": 0.4979848861694336,
|
| 948 |
+
"rewards/margins": 2.247861385345459,
|
| 949 |
+
"rewards/margins_max": 3.1490349769592285,
|
| 950 |
+
"rewards/margins_min": 1.3466877937316895,
|
| 951 |
+
"rewards/margins_std": 1.2744518518447876,
|
| 952 |
+
"rewards/rejected": -1.7498763799667358,
|
| 953 |
+
"step": 420
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"dpo_losses": 0.18394428491592407,
|
| 957 |
+
"epoch": 1.21,
|
| 958 |
+
"grad_norm": 89.7451007324007,
|
| 959 |
+
"learning_rate": 3.7238771323626822e-06,
|
| 960 |
+
"logits/chosen": -2.5034220218658447,
|
| 961 |
+
"logits/rejected": -2.4015450477600098,
|
| 962 |
+
"logps/chosen": -339.3304748535156,
|
| 963 |
+
"logps/rejected": -436.32769775390625,
|
| 964 |
+
"loss": 0.3026,
|
| 965 |
+
"positive_losses": 0.09120301902294159,
|
| 966 |
+
"rewards/accuracies": 1.0,
|
| 967 |
+
"rewards/chosen": 0.4775872230529785,
|
| 968 |
+
"rewards/margins": 2.2062175273895264,
|
| 969 |
+
"rewards/margins_max": 2.9012348651885986,
|
| 970 |
+
"rewards/margins_min": 1.5112000703811646,
|
| 971 |
+
"rewards/margins_std": 0.9829031229019165,
|
| 972 |
+
"rewards/rejected": -1.7286304235458374,
|
| 973 |
+
"step": 430
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"dpo_losses": 0.18616007268428802,
|
| 977 |
+
"epoch": 1.24,
|
| 978 |
+
"grad_norm": 3.3128247371778987,
|
| 979 |
+
"learning_rate": 3.651744727766676e-06,
|
| 980 |
+
"logits/chosen": -2.5054969787597656,
|
| 981 |
+
"logits/rejected": -2.38596773147583,
|
| 982 |
+
"logps/chosen": -275.51483154296875,
|
| 983 |
+
"logps/rejected": -352.8285827636719,
|
| 984 |
+
"loss": 0.3883,
|
| 985 |
+
"positive_losses": 0.630480945110321,
|
| 986 |
+
"rewards/accuracies": 1.0,
|
| 987 |
+
"rewards/chosen": 0.39738765358924866,
|
| 988 |
+
"rewards/margins": 1.9772555828094482,
|
| 989 |
+
"rewards/margins_max": 2.764988422393799,
|
| 990 |
+
"rewards/margins_min": 1.1895228624343872,
|
| 991 |
+
"rewards/margins_std": 1.1140224933624268,
|
| 992 |
+
"rewards/rejected": -1.579867959022522,
|
| 993 |
+
"step": 440
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"dpo_losses": 0.16775104403495789,
|
| 997 |
+
"epoch": 1.27,
|
| 998 |
+
"grad_norm": 2.373995116616861,
|
| 999 |
+
"learning_rate": 3.57837385160529e-06,
|
| 1000 |
+
"logits/chosen": -2.431580066680908,
|
| 1001 |
+
"logits/rejected": -2.3474247455596924,
|
| 1002 |
+
"logps/chosen": -304.8125915527344,
|
| 1003 |
+
"logps/rejected": -460.9014587402344,
|
| 1004 |
+
"loss": 0.3246,
|
| 1005 |
+
"positive_losses": 2.069793224334717,
|
| 1006 |
+
"rewards/accuracies": 1.0,
|
| 1007 |
+
"rewards/chosen": 0.33361926674842834,
|
| 1008 |
+
"rewards/margins": 2.2662878036499023,
|
| 1009 |
+
"rewards/margins_max": 2.9109106063842773,
|
| 1010 |
+
"rewards/margins_min": 1.6216650009155273,
|
| 1011 |
+
"rewards/margins_std": 0.9116341471672058,
|
| 1012 |
+
"rewards/rejected": -1.9326684474945068,
|
| 1013 |
+
"step": 450
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"dpo_losses": 0.13223373889923096,
|
| 1017 |
+
"epoch": 1.3,
|
| 1018 |
+
"grad_norm": 43.7205383502172,
|
| 1019 |
+
"learning_rate": 3.503843399610941e-06,
|
| 1020 |
+
"logits/chosen": -2.416710615158081,
|
| 1021 |
+
"logits/rejected": -2.4022932052612305,
|
| 1022 |
+
"logps/chosen": -355.0602111816406,
|
| 1023 |
+
"logps/rejected": -625.78369140625,
|
| 1024 |
+
"loss": 0.3215,
|
| 1025 |
+
"positive_losses": 3.091512680053711,
|
| 1026 |
+
"rewards/accuracies": 1.0,
|
| 1027 |
+
"rewards/chosen": 0.40217727422714233,
|
| 1028 |
+
"rewards/margins": 2.6340858936309814,
|
| 1029 |
+
"rewards/margins_max": 3.5210165977478027,
|
| 1030 |
+
"rewards/margins_min": 1.7471544742584229,
|
| 1031 |
+
"rewards/margins_std": 1.2543103694915771,
|
| 1032 |
+
"rewards/rejected": -2.2319083213806152,
|
| 1033 |
+
"step": 460
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"dpo_losses": 0.13370443880558014,
|
| 1037 |
+
"epoch": 1.32,
|
| 1038 |
+
"grad_norm": 93.10254553112391,
|
| 1039 |
+
"learning_rate": 3.4282335144083985e-06,
|
| 1040 |
+
"logits/chosen": -2.4071993827819824,
|
| 1041 |
+
"logits/rejected": -2.301910638809204,
|
| 1042 |
+
"logps/chosen": -323.6948547363281,
|
| 1043 |
+
"logps/rejected": -422.40106201171875,
|
| 1044 |
+
"loss": 0.3844,
|
| 1045 |
+
"positive_losses": 0.8020246624946594,
|
| 1046 |
+
"rewards/accuracies": 1.0,
|
| 1047 |
+
"rewards/chosen": 0.33555978536605835,
|
| 1048 |
+
"rewards/margins": 2.352768659591675,
|
| 1049 |
+
"rewards/margins_max": 2.6321401596069336,
|
| 1050 |
+
"rewards/margins_min": 2.073397159576416,
|
| 1051 |
+
"rewards/margins_std": 0.395090788602829,
|
| 1052 |
+
"rewards/rejected": -2.017209053039551,
|
| 1053 |
+
"step": 470
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"dpo_losses": 0.16256804764270782,
|
| 1057 |
+
"epoch": 1.35,
|
| 1058 |
+
"grad_norm": 25.781294981021368,
|
| 1059 |
+
"learning_rate": 3.351625499337395e-06,
|
| 1060 |
+
"logits/chosen": -2.5639257431030273,
|
| 1061 |
+
"logits/rejected": -2.3825831413269043,
|
| 1062 |
+
"logps/chosen": -368.14764404296875,
|
| 1063 |
+
"logps/rejected": -455.31170654296875,
|
| 1064 |
+
"loss": 0.3346,
|
| 1065 |
+
"positive_losses": 5.322105407714844,
|
| 1066 |
+
"rewards/accuracies": 1.0,
|
| 1067 |
+
"rewards/chosen": 0.47703951597213745,
|
| 1068 |
+
"rewards/margins": 2.266021251678467,
|
| 1069 |
+
"rewards/margins_max": 2.826010227203369,
|
| 1070 |
+
"rewards/margins_min": 1.706032156944275,
|
| 1071 |
+
"rewards/margins_std": 0.791944146156311,
|
| 1072 |
+
"rewards/rejected": -1.7889817953109741,
|
| 1073 |
+
"step": 480
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"dpo_losses": 0.27087026834487915,
|
| 1077 |
+
"epoch": 1.38,
|
| 1078 |
+
"grad_norm": 1.9694210301585493,
|
| 1079 |
+
"learning_rate": 3.2741017310271056e-06,
|
| 1080 |
+
"logits/chosen": -2.434044361114502,
|
| 1081 |
+
"logits/rejected": -2.2838988304138184,
|
| 1082 |
+
"logps/chosen": -185.84713745117188,
|
| 1083 |
+
"logps/rejected": -353.2933044433594,
|
| 1084 |
+
"loss": 0.4342,
|
| 1085 |
+
"positive_losses": 2.383211851119995,
|
| 1086 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1087 |
+
"rewards/chosen": 0.31434357166290283,
|
| 1088 |
+
"rewards/margins": 1.6815185546875,
|
| 1089 |
+
"rewards/margins_max": 2.3663582801818848,
|
| 1090 |
+
"rewards/margins_min": 0.9966787099838257,
|
| 1091 |
+
"rewards/margins_std": 0.9685096740722656,
|
| 1092 |
+
"rewards/rejected": -1.3671748638153076,
|
| 1093 |
+
"step": 490
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"dpo_losses": 0.1878516972064972,
|
| 1097 |
+
"epoch": 1.41,
|
| 1098 |
+
"grad_norm": 37.46970781719882,
|
| 1099 |
+
"learning_rate": 3.195745570816532e-06,
|
| 1100 |
+
"logits/chosen": -2.3778724670410156,
|
| 1101 |
+
"logits/rejected": -2.298856258392334,
|
| 1102 |
+
"logps/chosen": -339.0552978515625,
|
| 1103 |
+
"logps/rejected": -400.7374572753906,
|
| 1104 |
+
"loss": 0.3438,
|
| 1105 |
+
"positive_losses": 1.1473572254180908,
|
| 1106 |
+
"rewards/accuracies": 1.0,
|
| 1107 |
+
"rewards/chosen": 0.4514932632446289,
|
| 1108 |
+
"rewards/margins": 2.0797715187072754,
|
| 1109 |
+
"rewards/margins_max": 2.5511765480041504,
|
| 1110 |
+
"rewards/margins_min": 1.6083663702011108,
|
| 1111 |
+
"rewards/margins_std": 0.6666676998138428,
|
| 1112 |
+
"rewards/rejected": -1.628278136253357,
|
| 1113 |
+
"step": 500
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 1.41,
|
| 1117 |
+
"eval_dpo_losses": 0.6380993127822876,
|
| 1118 |
+
"eval_logits/chosen": -2.490675449371338,
|
| 1119 |
+
"eval_logits/rejected": -2.4471161365509033,
|
| 1120 |
+
"eval_logps/chosen": -385.29376220703125,
|
| 1121 |
+
"eval_logps/rejected": -393.24566650390625,
|
| 1122 |
+
"eval_loss": 10.641342163085938,
|
| 1123 |
+
"eval_positive_losses": 101.35254669189453,
|
| 1124 |
+
"eval_rewards/accuracies": 0.6865079402923584,
|
| 1125 |
+
"eval_rewards/chosen": -1.0007256269454956,
|
| 1126 |
+
"eval_rewards/margins": 0.33990418910980225,
|
| 1127 |
+
"eval_rewards/margins_max": 1.3353358507156372,
|
| 1128 |
+
"eval_rewards/margins_min": -0.6337663531303406,
|
| 1129 |
+
"eval_rewards/margins_std": 0.8805232644081116,
|
| 1130 |
+
"eval_rewards/rejected": -1.3406296968460083,
|
| 1131 |
+
"eval_runtime": 281.3613,
|
| 1132 |
+
"eval_samples_per_second": 7.108,
|
| 1133 |
+
"eval_steps_per_second": 0.224,
|
| 1134 |
+
"step": 500
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"dpo_losses": 0.15203723311424255,
|
| 1138 |
+
"epoch": 1.44,
|
| 1139 |
+
"grad_norm": 7.547244740150183,
|
| 1140 |
+
"learning_rate": 3.116641275116018e-06,
|
| 1141 |
+
"logits/chosen": -2.2004570960998535,
|
| 1142 |
+
"logits/rejected": -2.1605918407440186,
|
| 1143 |
+
"logps/chosen": -256.17730712890625,
|
| 1144 |
+
"logps/rejected": -509.5887756347656,
|
| 1145 |
+
"loss": 0.3429,
|
| 1146 |
+
"positive_losses": 2.6612415313720703,
|
| 1147 |
+
"rewards/accuracies": 1.0,
|
| 1148 |
+
"rewards/chosen": 0.3650303781032562,
|
| 1149 |
+
"rewards/margins": 2.308377504348755,
|
| 1150 |
+
"rewards/margins_max": 3.1350021362304688,
|
| 1151 |
+
"rewards/margins_min": 1.481752634048462,
|
| 1152 |
+
"rewards/margins_std": 1.1690237522125244,
|
| 1153 |
+
"rewards/rejected": -1.9433467388153076,
|
| 1154 |
+
"step": 510
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"dpo_losses": 0.12489266693592072,
|
| 1158 |
+
"epoch": 1.46,
|
| 1159 |
+
"grad_norm": 145.8887364997723,
|
| 1160 |
+
"learning_rate": 3.0368739048062956e-06,
|
| 1161 |
+
"logits/chosen": -2.448385238647461,
|
| 1162 |
+
"logits/rejected": -2.354788064956665,
|
| 1163 |
+
"logps/chosen": -307.94976806640625,
|
| 1164 |
+
"logps/rejected": -450.015380859375,
|
| 1165 |
+
"loss": 0.5303,
|
| 1166 |
+
"positive_losses": 3.1661620140075684,
|
| 1167 |
+
"rewards/accuracies": 1.0,
|
| 1168 |
+
"rewards/chosen": 0.3366628587245941,
|
| 1169 |
+
"rewards/margins": 2.504631519317627,
|
| 1170 |
+
"rewards/margins_max": 3.252946138381958,
|
| 1171 |
+
"rewards/margins_min": 1.7563165426254272,
|
| 1172 |
+
"rewards/margins_std": 1.058276891708374,
|
| 1173 |
+
"rewards/rejected": -2.16796875,
|
| 1174 |
+
"step": 520
|
| 1175 |
+
},
|
| 1176 |
+
{
|
| 1177 |
+
"dpo_losses": 0.14785408973693848,
|
| 1178 |
+
"epoch": 1.49,
|
| 1179 |
+
"grad_norm": 19.752652919627817,
|
| 1180 |
+
"learning_rate": 2.956529233772492e-06,
|
| 1181 |
+
"logits/chosen": -2.3587186336517334,
|
| 1182 |
+
"logits/rejected": -2.3671178817749023,
|
| 1183 |
+
"logps/chosen": -339.9403076171875,
|
| 1184 |
+
"logps/rejected": -447.1279296875,
|
| 1185 |
+
"loss": 0.2974,
|
| 1186 |
+
"positive_losses": 0.8550773859024048,
|
| 1187 |
+
"rewards/accuracies": 1.0,
|
| 1188 |
+
"rewards/chosen": 0.39065319299697876,
|
| 1189 |
+
"rewards/margins": 2.165048122406006,
|
| 1190 |
+
"rewards/margins_max": 2.7594637870788574,
|
| 1191 |
+
"rewards/margins_min": 1.5706324577331543,
|
| 1192 |
+
"rewards/margins_std": 0.8406306505203247,
|
| 1193 |
+
"rewards/rejected": -1.7743949890136719,
|
| 1194 |
+
"step": 530
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"dpo_losses": 0.1259264200925827,
|
| 1198 |
+
"epoch": 1.52,
|
| 1199 |
+
"grad_norm": 75.31648718698649,
|
| 1200 |
+
"learning_rate": 2.8756936566714317e-06,
|
| 1201 |
+
"logits/chosen": -2.5206494331359863,
|
| 1202 |
+
"logits/rejected": -2.379965305328369,
|
| 1203 |
+
"logps/chosen": -351.1200256347656,
|
| 1204 |
+
"logps/rejected": -426.8377990722656,
|
| 1205 |
+
"loss": 0.4729,
|
| 1206 |
+
"positive_losses": 1.8278591632843018,
|
| 1207 |
+
"rewards/accuracies": 1.0,
|
| 1208 |
+
"rewards/chosen": 0.48433223366737366,
|
| 1209 |
+
"rewards/margins": 2.3780269622802734,
|
| 1210 |
+
"rewards/margins_max": 2.926578998565674,
|
| 1211 |
+
"rewards/margins_min": 1.8294748067855835,
|
| 1212 |
+
"rewards/margins_std": 0.7757696509361267,
|
| 1213 |
+
"rewards/rejected": -1.893694281578064,
|
| 1214 |
+
"step": 540
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"dpo_losses": 0.20844757556915283,
|
| 1218 |
+
"epoch": 1.55,
|
| 1219 |
+
"grad_norm": 3.349676851087858,
|
| 1220 |
+
"learning_rate": 2.794454096031429e-06,
|
| 1221 |
+
"logits/chosen": -2.4806084632873535,
|
| 1222 |
+
"logits/rejected": -2.4055755138397217,
|
| 1223 |
+
"logps/chosen": -274.53900146484375,
|
| 1224 |
+
"logps/rejected": -464.46307373046875,
|
| 1225 |
+
"loss": 0.2124,
|
| 1226 |
+
"positive_losses": 1.989834189414978,
|
| 1227 |
+
"rewards/accuracies": 1.0,
|
| 1228 |
+
"rewards/chosen": 0.36409634351730347,
|
| 1229 |
+
"rewards/margins": 2.338207960128784,
|
| 1230 |
+
"rewards/margins_max": 3.5889182090759277,
|
| 1231 |
+
"rewards/margins_min": 1.0874969959259033,
|
| 1232 |
+
"rewards/margins_std": 1.7687723636627197,
|
| 1233 |
+
"rewards/rejected": -1.974111557006836,
|
| 1234 |
+
"step": 550
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"dpo_losses": 0.18082240223884583,
|
| 1238 |
+
"epoch": 1.58,
|
| 1239 |
+
"grad_norm": 2.0680076581049485,
|
| 1240 |
+
"learning_rate": 2.71289790878446e-06,
|
| 1241 |
+
"logits/chosen": -2.375046730041504,
|
| 1242 |
+
"logits/rejected": -2.316608190536499,
|
| 1243 |
+
"logps/chosen": -295.595703125,
|
| 1244 |
+
"logps/rejected": -556.7249755859375,
|
| 1245 |
+
"loss": 0.3351,
|
| 1246 |
+
"positive_losses": 4.852077484130859,
|
| 1247 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1248 |
+
"rewards/chosen": 0.21087315678596497,
|
| 1249 |
+
"rewards/margins": 2.390120267868042,
|
| 1250 |
+
"rewards/margins_max": 3.270312786102295,
|
| 1251 |
+
"rewards/margins_min": 1.5099279880523682,
|
| 1252 |
+
"rewards/margins_std": 1.24478018283844,
|
| 1253 |
+
"rewards/rejected": -2.1792471408843994,
|
| 1254 |
+
"step": 560
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"dpo_losses": 0.11622228473424911,
|
| 1258 |
+
"epoch": 1.61,
|
| 1259 |
+
"grad_norm": 53.09814348729035,
|
| 1260 |
+
"learning_rate": 2.6311127923312156e-06,
|
| 1261 |
+
"logits/chosen": -2.38314151763916,
|
| 1262 |
+
"logits/rejected": -2.2578694820404053,
|
| 1263 |
+
"logps/chosen": -386.07366943359375,
|
| 1264 |
+
"logps/rejected": -540.0533447265625,
|
| 1265 |
+
"loss": 0.2485,
|
| 1266 |
+
"positive_losses": 0.0,
|
| 1267 |
+
"rewards/accuracies": 1.0,
|
| 1268 |
+
"rewards/chosen": 0.4612743854522705,
|
| 1269 |
+
"rewards/margins": 2.703238010406494,
|
| 1270 |
+
"rewards/margins_max": 3.3717334270477295,
|
| 1271 |
+
"rewards/margins_min": 2.0347423553466797,
|
| 1272 |
+
"rewards/margins_std": 0.9453955888748169,
|
| 1273 |
+
"rewards/rejected": -2.2419633865356445,
|
| 1274 |
+
"step": 570
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"dpo_losses": 0.11324157565832138,
|
| 1278 |
+
"epoch": 1.63,
|
| 1279 |
+
"grad_norm": 28.84783476798183,
|
| 1280 |
+
"learning_rate": 2.549186690240057e-06,
|
| 1281 |
+
"logits/chosen": -2.3538641929626465,
|
| 1282 |
+
"logits/rejected": -2.2661383152008057,
|
| 1283 |
+
"logps/chosen": -265.77618408203125,
|
| 1284 |
+
"logps/rejected": -461.02362060546875,
|
| 1285 |
+
"loss": 0.3852,
|
| 1286 |
+
"positive_losses": 0.5109559893608093,
|
| 1287 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1288 |
+
"rewards/chosen": 0.326893150806427,
|
| 1289 |
+
"rewards/margins": 2.722642660140991,
|
| 1290 |
+
"rewards/margins_max": 3.37109375,
|
| 1291 |
+
"rewards/margins_min": 2.0741915702819824,
|
| 1292 |
+
"rewards/margins_std": 0.9170483350753784,
|
| 1293 |
+
"rewards/rejected": -2.395749568939209,
|
| 1294 |
+
"step": 580
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"dpo_losses": 0.19121481478214264,
|
| 1298 |
+
"epoch": 1.66,
|
| 1299 |
+
"grad_norm": 3.235430116905917,
|
| 1300 |
+
"learning_rate": 2.4672076976812548e-06,
|
| 1301 |
+
"logits/chosen": -2.256422281265259,
|
| 1302 |
+
"logits/rejected": -2.129117965698242,
|
| 1303 |
+
"logps/chosen": -335.0826721191406,
|
| 1304 |
+
"logps/rejected": -496.73138427734375,
|
| 1305 |
+
"loss": 0.3539,
|
| 1306 |
+
"positive_losses": 0.32223206758499146,
|
| 1307 |
+
"rewards/accuracies": 1.0,
|
| 1308 |
+
"rewards/chosen": 0.3417799472808838,
|
| 1309 |
+
"rewards/margins": 2.386916399002075,
|
| 1310 |
+
"rewards/margins_max": 3.4707369804382324,
|
| 1311 |
+
"rewards/margins_min": 1.3030953407287598,
|
| 1312 |
+
"rewards/margins_std": 1.5327543020248413,
|
| 1313 |
+
"rewards/rejected": -2.0451362133026123,
|
| 1314 |
+
"step": 590
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"dpo_losses": 0.2605344355106354,
|
| 1318 |
+
"epoch": 1.69,
|
| 1319 |
+
"grad_norm": 13.287666189135933,
|
| 1320 |
+
"learning_rate": 2.3852639666982218e-06,
|
| 1321 |
+
"logits/chosen": -2.4083638191223145,
|
| 1322 |
+
"logits/rejected": -2.3480491638183594,
|
| 1323 |
+
"logps/chosen": -243.49948120117188,
|
| 1324 |
+
"logps/rejected": -401.8130798339844,
|
| 1325 |
+
"loss": 0.2144,
|
| 1326 |
+
"positive_losses": 0.026648616418242455,
|
| 1327 |
+
"rewards/accuracies": 1.0,
|
| 1328 |
+
"rewards/chosen": 0.35153132677078247,
|
| 1329 |
+
"rewards/margins": 1.764525055885315,
|
| 1330 |
+
"rewards/margins_max": 2.218550205230713,
|
| 1331 |
+
"rewards/margins_min": 1.310499668121338,
|
| 1332 |
+
"rewards/margins_std": 0.642088770866394,
|
| 1333 |
+
"rewards/rejected": -1.4129936695098877,
|
| 1334 |
+
"step": 600
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 1.69,
|
| 1338 |
+
"eval_dpo_losses": 0.6267468929290771,
|
| 1339 |
+
"eval_logits/chosen": -2.3438363075256348,
|
| 1340 |
+
"eval_logits/rejected": -2.302872896194458,
|
| 1341 |
+
"eval_logps/chosen": -363.3914489746094,
|
| 1342 |
+
"eval_logps/rejected": -380.5317687988281,
|
| 1343 |
+
"eval_loss": 8.589625358581543,
|
| 1344 |
+
"eval_positive_losses": 79.79975891113281,
|
| 1345 |
+
"eval_rewards/accuracies": 0.6865079402923584,
|
| 1346 |
+
"eval_rewards/chosen": -0.7817028760910034,
|
| 1347 |
+
"eval_rewards/margins": 0.4317886233329773,
|
| 1348 |
+
"eval_rewards/margins_max": 1.5951305627822876,
|
| 1349 |
+
"eval_rewards/margins_min": -0.6660595536231995,
|
| 1350 |
+
"eval_rewards/margins_std": 1.0046793222427368,
|
| 1351 |
+
"eval_rewards/rejected": -1.2134915590286255,
|
| 1352 |
+
"eval_runtime": 280.6736,
|
| 1353 |
+
"eval_samples_per_second": 7.126,
|
| 1354 |
+
"eval_steps_per_second": 0.224,
|
| 1355 |
+
"step": 600
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"dpo_losses": 0.21777740120887756,
|
| 1359 |
+
"epoch": 1.72,
|
| 1360 |
+
"grad_norm": 3.6630257967936934,
|
| 1361 |
+
"learning_rate": 2.303443611417584e-06,
|
| 1362 |
+
"logits/chosen": -2.2210001945495605,
|
| 1363 |
+
"logits/rejected": -2.135100841522217,
|
| 1364 |
+
"logps/chosen": -324.7200622558594,
|
| 1365 |
+
"logps/rejected": -427.8872985839844,
|
| 1366 |
+
"loss": 0.4399,
|
| 1367 |
+
"positive_losses": 1.8486969470977783,
|
| 1368 |
+
"rewards/accuracies": 1.0,
|
| 1369 |
+
"rewards/chosen": 0.4076939523220062,
|
| 1370 |
+
"rewards/margins": 2.0033507347106934,
|
| 1371 |
+
"rewards/margins_max": 2.8805928230285645,
|
| 1372 |
+
"rewards/margins_min": 1.12610924243927,
|
| 1373 |
+
"rewards/margins_std": 1.2406072616577148,
|
| 1374 |
+
"rewards/rejected": -1.5956569910049438,
|
| 1375 |
+
"step": 610
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"dpo_losses": 0.16951796412467957,
|
| 1379 |
+
"epoch": 1.75,
|
| 1380 |
+
"grad_norm": 171.16064616866518,
|
| 1381 |
+
"learning_rate": 2.2218346133000264e-06,
|
| 1382 |
+
"logits/chosen": -2.2120890617370605,
|
| 1383 |
+
"logits/rejected": -2.1098124980926514,
|
| 1384 |
+
"logps/chosen": -283.6782531738281,
|
| 1385 |
+
"logps/rejected": -384.42193603515625,
|
| 1386 |
+
"loss": 0.445,
|
| 1387 |
+
"positive_losses": 6.724704742431641,
|
| 1388 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1389 |
+
"rewards/chosen": 0.2760510742664337,
|
| 1390 |
+
"rewards/margins": 2.14823317527771,
|
| 1391 |
+
"rewards/margins_max": 2.824097156524658,
|
| 1392 |
+
"rewards/margins_min": 1.4723690748214722,
|
| 1393 |
+
"rewards/margins_std": 0.9558159708976746,
|
| 1394 |
+
"rewards/rejected": -1.8721821308135986,
|
| 1395 |
+
"step": 620
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"dpo_losses": 0.15758368372917175,
|
| 1399 |
+
"epoch": 1.77,
|
| 1400 |
+
"grad_norm": 24.7162522936105,
|
| 1401 |
+
"learning_rate": 2.140524726533792e-06,
|
| 1402 |
+
"logits/chosen": -2.237668514251709,
|
| 1403 |
+
"logits/rejected": -2.1284663677215576,
|
| 1404 |
+
"logps/chosen": -338.96942138671875,
|
| 1405 |
+
"logps/rejected": -402.83697509765625,
|
| 1406 |
+
"loss": 0.3769,
|
| 1407 |
+
"positive_losses": 0.14499235153198242,
|
| 1408 |
+
"rewards/accuracies": 1.0,
|
| 1409 |
+
"rewards/chosen": 0.5579421520233154,
|
| 1410 |
+
"rewards/margins": 2.344921350479126,
|
| 1411 |
+
"rewards/margins_max": 2.9339561462402344,
|
| 1412 |
+
"rewards/margins_min": 1.7558867931365967,
|
| 1413 |
+
"rewards/margins_std": 0.833020806312561,
|
| 1414 |
+
"rewards/rejected": -1.7869793176651,
|
| 1415 |
+
"step": 630
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"dpo_losses": 0.18143755197525024,
|
| 1419 |
+
"epoch": 1.8,
|
| 1420 |
+
"grad_norm": 28.54900764303978,
|
| 1421 |
+
"learning_rate": 2.059601383672566e-06,
|
| 1422 |
+
"logits/chosen": -2.421058177947998,
|
| 1423 |
+
"logits/rejected": -2.359043598175049,
|
| 1424 |
+
"logps/chosen": -261.0716857910156,
|
| 1425 |
+
"logps/rejected": -371.5032653808594,
|
| 1426 |
+
"loss": 0.2764,
|
| 1427 |
+
"positive_losses": 0.5634332895278931,
|
| 1428 |
+
"rewards/accuracies": 1.0,
|
| 1429 |
+
"rewards/chosen": 0.4140376150608063,
|
| 1430 |
+
"rewards/margins": 2.1301121711730957,
|
| 1431 |
+
"rewards/margins_max": 2.458400011062622,
|
| 1432 |
+
"rewards/margins_min": 1.8018243312835693,
|
| 1433 |
+
"rewards/margins_std": 0.46426907181739807,
|
| 1434 |
+
"rewards/rejected": -1.7160745859146118,
|
| 1435 |
+
"step": 640
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"dpo_losses": 0.21564999222755432,
|
| 1439 |
+
"epoch": 1.83,
|
| 1440 |
+
"grad_norm": 43.347219579951144,
|
| 1441 |
+
"learning_rate": 1.9791516016192214e-06,
|
| 1442 |
+
"logits/chosen": -2.4876837730407715,
|
| 1443 |
+
"logits/rejected": -2.327549457550049,
|
| 1444 |
+
"logps/chosen": -288.0809020996094,
|
| 1445 |
+
"logps/rejected": -400.9809875488281,
|
| 1446 |
+
"loss": 0.2601,
|
| 1447 |
+
"positive_losses": 2.1945955753326416,
|
| 1448 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1449 |
+
"rewards/chosen": 0.4631883203983307,
|
| 1450 |
+
"rewards/margins": 2.244870662689209,
|
| 1451 |
+
"rewards/margins_max": 3.217092990875244,
|
| 1452 |
+
"rewards/margins_min": 1.2726480960845947,
|
| 1453 |
+
"rewards/margins_std": 1.3749301433563232,
|
| 1454 |
+
"rewards/rejected": -1.7816823720932007,
|
| 1455 |
+
"step": 650
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"dpo_losses": 0.21185970306396484,
|
| 1459 |
+
"epoch": 1.86,
|
| 1460 |
+
"grad_norm": 4.895621870086889,
|
| 1461 |
+
"learning_rate": 1.8992618880565039e-06,
|
| 1462 |
+
"logits/chosen": -2.147622585296631,
|
| 1463 |
+
"logits/rejected": -2.0906789302825928,
|
| 1464 |
+
"logps/chosen": -282.22711181640625,
|
| 1465 |
+
"logps/rejected": -362.41119384765625,
|
| 1466 |
+
"loss": 0.4916,
|
| 1467 |
+
"positive_losses": 3.9535961151123047,
|
| 1468 |
+
"rewards/accuracies": 1.0,
|
| 1469 |
+
"rewards/chosen": 0.4179634153842926,
|
| 1470 |
+
"rewards/margins": 2.141587495803833,
|
| 1471 |
+
"rewards/margins_max": 3.0650925636291504,
|
| 1472 |
+
"rewards/margins_min": 1.218082308769226,
|
| 1473 |
+
"rewards/margins_std": 1.3060333728790283,
|
| 1474 |
+
"rewards/rejected": -1.7236239910125732,
|
| 1475 |
+
"step": 660
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"dpo_losses": 0.11567674577236176,
|
| 1479 |
+
"epoch": 1.89,
|
| 1480 |
+
"grad_norm": 259.9155211313861,
|
| 1481 |
+
"learning_rate": 1.8200181484252888e-06,
|
| 1482 |
+
"logits/chosen": -2.390129566192627,
|
| 1483 |
+
"logits/rejected": -2.4018349647521973,
|
| 1484 |
+
"logps/chosen": -326.493896484375,
|
| 1485 |
+
"logps/rejected": -549.7818603515625,
|
| 1486 |
+
"loss": 0.2561,
|
| 1487 |
+
"positive_losses": 0.0,
|
| 1488 |
+
"rewards/accuracies": 1.0,
|
| 1489 |
+
"rewards/chosen": 0.562630295753479,
|
| 1490 |
+
"rewards/margins": 2.886990785598755,
|
| 1491 |
+
"rewards/margins_max": 3.560823440551758,
|
| 1492 |
+
"rewards/margins_min": 2.2131576538085938,
|
| 1493 |
+
"rewards/margins_std": 0.9529436826705933,
|
| 1494 |
+
"rewards/rejected": -2.3243603706359863,
|
| 1495 |
+
"step": 670
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"dpo_losses": 0.1450214684009552,
|
| 1499 |
+
"epoch": 1.92,
|
| 1500 |
+
"grad_norm": 3.1779421953498246,
|
| 1501 |
+
"learning_rate": 1.7415055935504234e-06,
|
| 1502 |
+
"logits/chosen": -2.4584078788757324,
|
| 1503 |
+
"logits/rejected": -2.2816786766052246,
|
| 1504 |
+
"logps/chosen": -351.3116760253906,
|
| 1505 |
+
"logps/rejected": -523.8319091796875,
|
| 1506 |
+
"loss": 0.4032,
|
| 1507 |
+
"positive_losses": 3.919013500213623,
|
| 1508 |
+
"rewards/accuracies": 1.0,
|
| 1509 |
+
"rewards/chosen": 0.4204772114753723,
|
| 1510 |
+
"rewards/margins": 2.6883883476257324,
|
| 1511 |
+
"rewards/margins_max": 3.6081409454345703,
|
| 1512 |
+
"rewards/margins_min": 1.7686359882354736,
|
| 1513 |
+
"rewards/margins_std": 1.300726294517517,
|
| 1514 |
+
"rewards/rejected": -2.267911434173584,
|
| 1515 |
+
"step": 680
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"dpo_losses": 0.22765210270881653,
|
| 1519 |
+
"epoch": 1.94,
|
| 1520 |
+
"grad_norm": 290.4438144891366,
|
| 1521 |
+
"learning_rate": 1.6638086480134954e-06,
|
| 1522 |
+
"logits/chosen": -2.2493913173675537,
|
| 1523 |
+
"logits/rejected": -2.2027366161346436,
|
| 1524 |
+
"logps/chosen": -208.2427520751953,
|
| 1525 |
+
"logps/rejected": -294.7981872558594,
|
| 1526 |
+
"loss": 0.3929,
|
| 1527 |
+
"positive_losses": 4.185807228088379,
|
| 1528 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1529 |
+
"rewards/chosen": 0.2872256338596344,
|
| 1530 |
+
"rewards/margins": 1.9401639699935913,
|
| 1531 |
+
"rewards/margins_max": 2.823594331741333,
|
| 1532 |
+
"rewards/margins_min": 1.0567338466644287,
|
| 1533 |
+
"rewards/margins_std": 1.249358892440796,
|
| 1534 |
+
"rewards/rejected": -1.6529382467269897,
|
| 1535 |
+
"step": 690
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"dpo_losses": 0.11677996069192886,
|
| 1539 |
+
"epoch": 1.97,
|
| 1540 |
+
"grad_norm": 2.93724058913164,
|
| 1541 |
+
"learning_rate": 1.5870108593710473e-06,
|
| 1542 |
+
"logits/chosen": -2.169877290725708,
|
| 1543 |
+
"logits/rejected": -2.062708854675293,
|
| 1544 |
+
"logps/chosen": -379.1634826660156,
|
| 1545 |
+
"logps/rejected": -440.5103454589844,
|
| 1546 |
+
"loss": 0.3314,
|
| 1547 |
+
"positive_losses": 0.9755552411079407,
|
| 1548 |
+
"rewards/accuracies": 1.0,
|
| 1549 |
+
"rewards/chosen": 0.478015273809433,
|
| 1550 |
+
"rewards/margins": 2.7329649925231934,
|
| 1551 |
+
"rewards/margins_max": 3.358954906463623,
|
| 1552 |
+
"rewards/margins_min": 2.1069746017456055,
|
| 1553 |
+
"rewards/margins_std": 0.8852837681770325,
|
| 1554 |
+
"rewards/rejected": -2.2549493312835693,
|
| 1555 |
+
"step": 700
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 1.97,
|
| 1559 |
+
"eval_dpo_losses": 0.6525446176528931,
|
| 1560 |
+
"eval_logits/chosen": -2.343222141265869,
|
| 1561 |
+
"eval_logits/rejected": -2.30248761177063,
|
| 1562 |
+
"eval_logps/chosen": -391.17047119140625,
|
| 1563 |
+
"eval_logps/rejected": -410.6868591308594,
|
| 1564 |
+
"eval_loss": 11.165140151977539,
|
| 1565 |
+
"eval_positive_losses": 107.29693603515625,
|
| 1566 |
+
"eval_rewards/accuracies": 0.6626983880996704,
|
| 1567 |
+
"eval_rewards/chosen": -1.0594924688339233,
|
| 1568 |
+
"eval_rewards/margins": 0.4555494785308838,
|
| 1569 |
+
"eval_rewards/margins_max": 1.7776461839675903,
|
| 1570 |
+
"eval_rewards/margins_min": -0.8450111150741577,
|
| 1571 |
+
"eval_rewards/margins_std": 1.16599440574646,
|
| 1572 |
+
"eval_rewards/rejected": -1.5150419473648071,
|
| 1573 |
+
"eval_runtime": 281.451,
|
| 1574 |
+
"eval_samples_per_second": 7.106,
|
| 1575 |
+
"eval_steps_per_second": 0.224,
|
| 1576 |
+
"step": 700
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"dpo_losses": 0.2008899748325348,
|
| 1580 |
+
"epoch": 2.0,
|
| 1581 |
+
"grad_norm": 1.5771255349778672,
|
| 1582 |
+
"learning_rate": 1.511194808315853e-06,
|
| 1583 |
+
"logits/chosen": -2.2691843509674072,
|
| 1584 |
+
"logits/rejected": -2.16029691696167,
|
| 1585 |
+
"logps/chosen": -264.2540588378906,
|
| 1586 |
+
"logps/rejected": -368.97308349609375,
|
| 1587 |
+
"loss": 0.2577,
|
| 1588 |
+
"positive_losses": 1.1838890314102173,
|
| 1589 |
+
"rewards/accuracies": 1.0,
|
| 1590 |
+
"rewards/chosen": 0.4201991558074951,
|
| 1591 |
+
"rewards/margins": 2.272444248199463,
|
| 1592 |
+
"rewards/margins_max": 3.1316885948181152,
|
| 1593 |
+
"rewards/margins_min": 1.413199782371521,
|
| 1594 |
+
"rewards/margins_std": 1.2151552438735962,
|
| 1595 |
+
"rewards/rejected": -1.8522450923919678,
|
| 1596 |
+
"step": 710
|
| 1597 |
+
},
|
| 1598 |
+
{
|
| 1599 |
+
"dpo_losses": 0.11133086681365967,
|
| 1600 |
+
"epoch": 2.03,
|
| 1601 |
+
"grad_norm": 39.920300238021134,
|
| 1602 |
+
"learning_rate": 1.4364420198778662e-06,
|
| 1603 |
+
"logits/chosen": -2.3695783615112305,
|
| 1604 |
+
"logits/rejected": -2.292491912841797,
|
| 1605 |
+
"logps/chosen": -303.01556396484375,
|
| 1606 |
+
"logps/rejected": -584.0543212890625,
|
| 1607 |
+
"loss": 0.2033,
|
| 1608 |
+
"positive_losses": 1.3720115423202515,
|
| 1609 |
+
"rewards/accuracies": 1.0,
|
| 1610 |
+
"rewards/chosen": 0.44599366188049316,
|
| 1611 |
+
"rewards/margins": 2.9826126098632812,
|
| 1612 |
+
"rewards/margins_max": 3.897780179977417,
|
| 1613 |
+
"rewards/margins_min": 2.0674448013305664,
|
| 1614 |
+
"rewards/margins_std": 1.2942426204681396,
|
| 1615 |
+
"rewards/rejected": -2.536618709564209,
|
| 1616 |
+
"step": 720
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"dpo_losses": 0.06813247501850128,
|
| 1620 |
+
"epoch": 2.06,
|
| 1621 |
+
"grad_norm": 61.35667528119903,
|
| 1622 |
+
"learning_rate": 1.3628328757603243e-06,
|
| 1623 |
+
"logits/chosen": -2.3125011920928955,
|
| 1624 |
+
"logits/rejected": -2.234978437423706,
|
| 1625 |
+
"logps/chosen": -380.5849609375,
|
| 1626 |
+
"logps/rejected": -546.2708129882812,
|
| 1627 |
+
"loss": 0.2931,
|
| 1628 |
+
"positive_losses": 3.713000535964966,
|
| 1629 |
+
"rewards/accuracies": 1.0,
|
| 1630 |
+
"rewards/chosen": 0.42165422439575195,
|
| 1631 |
+
"rewards/margins": 3.356724500656128,
|
| 1632 |
+
"rewards/margins_max": 3.9914348125457764,
|
| 1633 |
+
"rewards/margins_min": 2.7220141887664795,
|
| 1634 |
+
"rewards/margins_std": 0.8976157903671265,
|
| 1635 |
+
"rewards/rejected": -2.935070276260376,
|
| 1636 |
+
"step": 730
|
| 1637 |
+
},
|
| 1638 |
+
{
|
| 1639 |
+
"dpo_losses": 0.1471334546804428,
|
| 1640 |
+
"epoch": 2.08,
|
| 1641 |
+
"grad_norm": 81.02956274757072,
|
| 1642 |
+
"learning_rate": 1.2904465279052725e-06,
|
| 1643 |
+
"logits/chosen": -2.3687386512756348,
|
| 1644 |
+
"logits/rejected": -2.246112585067749,
|
| 1645 |
+
"logps/chosen": -307.13690185546875,
|
| 1646 |
+
"logps/rejected": -456.7256774902344,
|
| 1647 |
+
"loss": 0.2511,
|
| 1648 |
+
"positive_losses": 0.0,
|
| 1649 |
+
"rewards/accuracies": 1.0,
|
| 1650 |
+
"rewards/chosen": 0.44816774129867554,
|
| 1651 |
+
"rewards/margins": 2.7628419399261475,
|
| 1652 |
+
"rewards/margins_max": 3.8002758026123047,
|
| 1653 |
+
"rewards/margins_min": 1.7254081964492798,
|
| 1654 |
+
"rewards/margins_std": 1.4671531915664673,
|
| 1655 |
+
"rewards/rejected": -2.314674139022827,
|
| 1656 |
+
"step": 740
|
| 1657 |
+
},
|
| 1658 |
+
{
|
| 1659 |
+
"dpo_losses": 0.18885108828544617,
|
| 1660 |
+
"epoch": 2.11,
|
| 1661 |
+
"grad_norm": 37.34403765579587,
|
| 1662 |
+
"learning_rate": 1.219360813381446e-06,
|
| 1663 |
+
"logits/chosen": -2.1814396381378174,
|
| 1664 |
+
"logits/rejected": -2.133742570877075,
|
| 1665 |
+
"logps/chosen": -167.7379913330078,
|
| 1666 |
+
"logps/rejected": -342.445068359375,
|
| 1667 |
+
"loss": 0.2246,
|
| 1668 |
+
"positive_losses": 3.9296765327453613,
|
| 1669 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1670 |
+
"rewards/chosen": 0.2083771526813507,
|
| 1671 |
+
"rewards/margins": 2.266416072845459,
|
| 1672 |
+
"rewards/margins_max": 3.0310301780700684,
|
| 1673 |
+
"rewards/margins_min": 1.501802682876587,
|
| 1674 |
+
"rewards/margins_std": 1.0813268423080444,
|
| 1675 |
+
"rewards/rejected": -2.0580391883850098,
|
| 1676 |
+
"step": 750
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"dpo_losses": 0.07930545508861542,
|
| 1680 |
+
"epoch": 2.14,
|
| 1681 |
+
"grad_norm": 53.31665337719304,
|
| 1682 |
+
"learning_rate": 1.1496521706860392e-06,
|
| 1683 |
+
"logits/chosen": -2.3485515117645264,
|
| 1684 |
+
"logits/rejected": -2.2168877124786377,
|
| 1685 |
+
"logps/chosen": -298.58966064453125,
|
| 1686 |
+
"logps/rejected": -549.2097778320312,
|
| 1687 |
+
"loss": 0.2258,
|
| 1688 |
+
"positive_losses": 3.5301315784454346,
|
| 1689 |
+
"rewards/accuracies": 1.0,
|
| 1690 |
+
"rewards/chosen": 0.344715416431427,
|
| 1691 |
+
"rewards/margins": 3.2088115215301514,
|
| 1692 |
+
"rewards/margins_max": 4.005995750427246,
|
| 1693 |
+
"rewards/margins_min": 2.411628246307373,
|
| 1694 |
+
"rewards/margins_std": 1.1273881196975708,
|
| 1695 |
+
"rewards/rejected": -2.8640968799591064,
|
| 1696 |
+
"step": 760
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"dpo_losses": 0.13964664936065674,
|
| 1700 |
+
"epoch": 2.17,
|
| 1701 |
+
"grad_norm": 1.2653188452887754,
|
| 1702 |
+
"learning_rate": 1.0813955575503588e-06,
|
| 1703 |
+
"logits/chosen": -2.278029680252075,
|
| 1704 |
+
"logits/rejected": -2.2471253871917725,
|
| 1705 |
+
"logps/chosen": -296.67205810546875,
|
| 1706 |
+
"logps/rejected": -532.7224731445312,
|
| 1707 |
+
"loss": 0.2746,
|
| 1708 |
+
"positive_losses": 1.8862769603729248,
|
| 1709 |
+
"rewards/accuracies": 1.0,
|
| 1710 |
+
"rewards/chosen": 0.36964336037635803,
|
| 1711 |
+
"rewards/margins": 3.1202445030212402,
|
| 1712 |
+
"rewards/margins_max": 4.3073296546936035,
|
| 1713 |
+
"rewards/margins_min": 1.9331591129302979,
|
| 1714 |
+
"rewards/margins_std": 1.6787922382354736,
|
| 1715 |
+
"rewards/rejected": -2.750600814819336,
|
| 1716 |
+
"step": 770
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"dpo_losses": 0.18134805560112,
|
| 1720 |
+
"epoch": 2.2,
|
| 1721 |
+
"grad_norm": 49.5060636871573,
|
| 1722 |
+
"learning_rate": 1.0146643703377488e-06,
|
| 1723 |
+
"logits/chosen": -2.3735833168029785,
|
| 1724 |
+
"logits/rejected": -2.250998020172119,
|
| 1725 |
+
"logps/chosen": -292.6219177246094,
|
| 1726 |
+
"logps/rejected": -449.14068603515625,
|
| 1727 |
+
"loss": 0.2041,
|
| 1728 |
+
"positive_losses": 1.1141910552978516,
|
| 1729 |
+
"rewards/accuracies": 1.0,
|
| 1730 |
+
"rewards/chosen": 0.3442000448703766,
|
| 1731 |
+
"rewards/margins": 2.5554795265197754,
|
| 1732 |
+
"rewards/margins_max": 3.6242382526397705,
|
| 1733 |
+
"rewards/margins_min": 1.4867204427719116,
|
| 1734 |
+
"rewards/margins_std": 1.51145339012146,
|
| 1735 |
+
"rewards/rejected": -2.2112793922424316,
|
| 1736 |
+
"step": 780
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"dpo_losses": 0.09527120739221573,
|
| 1740 |
+
"epoch": 2.23,
|
| 1741 |
+
"grad_norm": 22.61766042791389,
|
| 1742 |
+
"learning_rate": 9.495303651204496e-07,
|
| 1743 |
+
"logits/chosen": -2.317281484603882,
|
| 1744 |
+
"logits/rejected": -2.2382774353027344,
|
| 1745 |
+
"logps/chosen": -320.7598571777344,
|
| 1746 |
+
"logps/rejected": -518.2537841796875,
|
| 1747 |
+
"loss": 0.1938,
|
| 1748 |
+
"positive_losses": 0.0011909485328942537,
|
| 1749 |
+
"rewards/accuracies": 1.0,
|
| 1750 |
+
"rewards/chosen": 0.4521896243095398,
|
| 1751 |
+
"rewards/margins": 2.8671340942382812,
|
| 1752 |
+
"rewards/margins_max": 3.515406847000122,
|
| 1753 |
+
"rewards/margins_min": 2.2188615798950195,
|
| 1754 |
+
"rewards/margins_std": 0.9167959094047546,
|
| 1755 |
+
"rewards/rejected": -2.414944648742676,
|
| 1756 |
+
"step": 790
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"dpo_losses": 0.09652134031057358,
|
| 1760 |
+
"epoch": 2.25,
|
| 1761 |
+
"grad_norm": 55.89571807239017,
|
| 1762 |
+
"learning_rate": 8.860635805202616e-07,
|
| 1763 |
+
"logits/chosen": -2.334362506866455,
|
| 1764 |
+
"logits/rejected": -2.2386062145233154,
|
| 1765 |
+
"logps/chosen": -323.18060302734375,
|
| 1766 |
+
"logps/rejected": -481.80712890625,
|
| 1767 |
+
"loss": 0.1352,
|
| 1768 |
+
"positive_losses": 0.3718675673007965,
|
| 1769 |
+
"rewards/accuracies": 1.0,
|
| 1770 |
+
"rewards/chosen": 0.4021781086921692,
|
| 1771 |
+
"rewards/margins": 2.7669034004211426,
|
| 1772 |
+
"rewards/margins_max": 3.376081943511963,
|
| 1773 |
+
"rewards/margins_min": 2.1577250957489014,
|
| 1774 |
+
"rewards/margins_std": 0.8615081906318665,
|
| 1775 |
+
"rewards/rejected": -2.364725112915039,
|
| 1776 |
+
"step": 800
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 2.25,
|
| 1780 |
+
"eval_dpo_losses": 0.6700040102005005,
|
| 1781 |
+
"eval_logits/chosen": -2.3320279121398926,
|
| 1782 |
+
"eval_logits/rejected": -2.290130376815796,
|
| 1783 |
+
"eval_logps/chosen": -415.07861328125,
|
| 1784 |
+
"eval_logps/rejected": -441.023681640625,
|
| 1785 |
+
"eval_loss": 13.35706615447998,
|
| 1786 |
+
"eval_positive_losses": 130.9070281982422,
|
| 1787 |
+
"eval_rewards/accuracies": 0.6626983880996704,
|
| 1788 |
+
"eval_rewards/chosen": -1.2985737323760986,
|
| 1789 |
+
"eval_rewards/margins": 0.5198364853858948,
|
| 1790 |
+
"eval_rewards/margins_max": 2.022522211074829,
|
| 1791 |
+
"eval_rewards/margins_min": -0.9602744579315186,
|
| 1792 |
+
"eval_rewards/margins_std": 1.3295913934707642,
|
| 1793 |
+
"eval_rewards/rejected": -1.8184101581573486,
|
| 1794 |
+
"eval_runtime": 281.6798,
|
| 1795 |
+
"eval_samples_per_second": 7.1,
|
| 1796 |
+
"eval_steps_per_second": 0.224,
|
| 1797 |
+
"step": 800
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"dpo_losses": 0.11107480525970459,
|
| 1801 |
+
"epoch": 2.28,
|
| 1802 |
+
"grad_norm": 1.211879634445009,
|
| 1803 |
+
"learning_rate": 8.24332262395994e-07,
|
| 1804 |
+
"logits/chosen": -2.4299230575561523,
|
| 1805 |
+
"logits/rejected": -2.3509979248046875,
|
| 1806 |
+
"logps/chosen": -259.64581298828125,
|
| 1807 |
+
"logps/rejected": -510.4208984375,
|
| 1808 |
+
"loss": 0.4972,
|
| 1809 |
+
"positive_losses": 1.13177490234375,
|
| 1810 |
+
"rewards/accuracies": 1.0,
|
| 1811 |
+
"rewards/chosen": 0.34921571612358093,
|
| 1812 |
+
"rewards/margins": 3.014543056488037,
|
| 1813 |
+
"rewards/margins_max": 4.297389984130859,
|
| 1814 |
+
"rewards/margins_min": 1.7316957712173462,
|
| 1815 |
+
"rewards/margins_std": 1.8142198324203491,
|
| 1816 |
+
"rewards/rejected": -2.665327310562134,
|
| 1817 |
+
"step": 810
|
| 1818 |
+
},
|
| 1819 |
+
{
|
| 1820 |
+
"dpo_losses": 0.15867747366428375,
|
| 1821 |
+
"epoch": 2.31,
|
| 1822 |
+
"grad_norm": 1.9974082008055192,
|
| 1823 |
+
"learning_rate": 7.644027904586587e-07,
|
| 1824 |
+
"logits/chosen": -2.368762731552124,
|
| 1825 |
+
"logits/rejected": -2.2775216102600098,
|
| 1826 |
+
"logps/chosen": -270.0508728027344,
|
| 1827 |
+
"logps/rejected": -446.8939514160156,
|
| 1828 |
+
"loss": 0.3191,
|
| 1829 |
+
"positive_losses": 0.852569580078125,
|
| 1830 |
+
"rewards/accuracies": 1.0,
|
| 1831 |
+
"rewards/chosen": 0.3547489047050476,
|
| 1832 |
+
"rewards/margins": 2.747641086578369,
|
| 1833 |
+
"rewards/margins_max": 3.639145612716675,
|
| 1834 |
+
"rewards/margins_min": 1.8561369180679321,
|
| 1835 |
+
"rewards/margins_std": 1.2607777118682861,
|
| 1836 |
+
"rewards/rejected": -2.392892360687256,
|
| 1837 |
+
"step": 820
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"dpo_losses": 0.06032683700323105,
|
| 1841 |
+
"epoch": 2.34,
|
| 1842 |
+
"grad_norm": 79.25640130890059,
|
| 1843 |
+
"learning_rate": 7.06339606893347e-07,
|
| 1844 |
+
"logits/chosen": -2.3674511909484863,
|
| 1845 |
+
"logits/rejected": -2.254099130630493,
|
| 1846 |
+
"logps/chosen": -410.2024841308594,
|
| 1847 |
+
"logps/rejected": -582.8692016601562,
|
| 1848 |
+
"loss": 0.212,
|
| 1849 |
+
"positive_losses": 3.602843761444092,
|
| 1850 |
+
"rewards/accuracies": 1.0,
|
| 1851 |
+
"rewards/chosen": 0.4369390904903412,
|
| 1852 |
+
"rewards/margins": 3.7233974933624268,
|
| 1853 |
+
"rewards/margins_max": 4.673032283782959,
|
| 1854 |
+
"rewards/margins_min": 2.7737619876861572,
|
| 1855 |
+
"rewards/margins_std": 1.3429871797561646,
|
| 1856 |
+
"rewards/rejected": -3.2864582538604736,
|
| 1857 |
+
"step": 830
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"dpo_losses": 0.12710335850715637,
|
| 1861 |
+
"epoch": 2.37,
|
| 1862 |
+
"grad_norm": 5.277843753781562,
|
| 1863 |
+
"learning_rate": 6.502051470645149e-07,
|
| 1864 |
+
"logits/chosen": -2.486243486404419,
|
| 1865 |
+
"logits/rejected": -2.29672908782959,
|
| 1866 |
+
"logps/chosen": -348.5108947753906,
|
| 1867 |
+
"logps/rejected": -525.3292846679688,
|
| 1868 |
+
"loss": 0.346,
|
| 1869 |
+
"positive_losses": 0.0202178955078125,
|
| 1870 |
+
"rewards/accuracies": 1.0,
|
| 1871 |
+
"rewards/chosen": 0.4295215606689453,
|
| 1872 |
+
"rewards/margins": 2.8306221961975098,
|
| 1873 |
+
"rewards/margins_max": 3.4442856311798096,
|
| 1874 |
+
"rewards/margins_min": 2.216958522796631,
|
| 1875 |
+
"rewards/margins_std": 0.8678513765335083,
|
| 1876 |
+
"rewards/rejected": -2.4011006355285645,
|
| 1877 |
+
"step": 840
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"dpo_losses": 0.10362281650304794,
|
| 1881 |
+
"epoch": 2.39,
|
| 1882 |
+
"grad_norm": 172.11522416865105,
|
| 1883 |
+
"learning_rate": 5.960597723792194e-07,
|
| 1884 |
+
"logits/chosen": -2.351360559463501,
|
| 1885 |
+
"logits/rejected": -2.166510581970215,
|
| 1886 |
+
"logps/chosen": -337.3572692871094,
|
| 1887 |
+
"logps/rejected": -561.3741455078125,
|
| 1888 |
+
"loss": 0.2227,
|
| 1889 |
+
"positive_losses": 0.215586856007576,
|
| 1890 |
+
"rewards/accuracies": 1.0,
|
| 1891 |
+
"rewards/chosen": 0.47708138823509216,
|
| 1892 |
+
"rewards/margins": 3.5102126598358154,
|
| 1893 |
+
"rewards/margins_max": 4.734680652618408,
|
| 1894 |
+
"rewards/margins_min": 2.28574538230896,
|
| 1895 |
+
"rewards/margins_std": 1.731658697128296,
|
| 1896 |
+
"rewards/rejected": -3.0331313610076904,
|
| 1897 |
+
"step": 850
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"dpo_losses": 0.10238895565271378,
|
| 1901 |
+
"epoch": 2.42,
|
| 1902 |
+
"grad_norm": 82.67868906193421,
|
| 1903 |
+
"learning_rate": 5.43961705380465e-07,
|
| 1904 |
+
"logits/chosen": -2.389751434326172,
|
| 1905 |
+
"logits/rejected": -2.2885704040527344,
|
| 1906 |
+
"logps/chosen": -315.6968078613281,
|
| 1907 |
+
"logps/rejected": -601.2724609375,
|
| 1908 |
+
"loss": 0.2321,
|
| 1909 |
+
"positive_losses": 0.0,
|
| 1910 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1911 |
+
"rewards/chosen": 0.3972366154193878,
|
| 1912 |
+
"rewards/margins": 3.726912260055542,
|
| 1913 |
+
"rewards/margins_max": 4.830050945281982,
|
| 1914 |
+
"rewards/margins_min": 2.6237740516662598,
|
| 1915 |
+
"rewards/margins_std": 1.5600733757019043,
|
| 1916 |
+
"rewards/rejected": -3.3296761512756348,
|
| 1917 |
+
"step": 860
|
| 1918 |
+
},
|
| 1919 |
+
{
|
| 1920 |
+
"dpo_losses": 0.10260417312383652,
|
| 1921 |
+
"epoch": 2.45,
|
| 1922 |
+
"grad_norm": 24.22347862103844,
|
| 1923 |
+
"learning_rate": 4.939669671404871e-07,
|
| 1924 |
+
"logits/chosen": -2.3038783073425293,
|
| 1925 |
+
"logits/rejected": -2.185800552368164,
|
| 1926 |
+
"logps/chosen": -305.2945556640625,
|
| 1927 |
+
"logps/rejected": -611.5335693359375,
|
| 1928 |
+
"loss": 0.1672,
|
| 1929 |
+
"positive_losses": 0.3565734922885895,
|
| 1930 |
+
"rewards/accuracies": 1.0,
|
| 1931 |
+
"rewards/chosen": 0.35259321331977844,
|
| 1932 |
+
"rewards/margins": 3.309342861175537,
|
| 1933 |
+
"rewards/margins_max": 4.305205345153809,
|
| 1934 |
+
"rewards/margins_min": 2.313480854034424,
|
| 1935 |
+
"rewards/margins_std": 1.4083621501922607,
|
| 1936 |
+
"rewards/rejected": -2.956749439239502,
|
| 1937 |
+
"step": 870
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"dpo_losses": 0.0972672775387764,
|
| 1941 |
+
"epoch": 2.48,
|
| 1942 |
+
"grad_norm": 2.5325654932498978,
|
| 1943 |
+
"learning_rate": 4.461293170212644e-07,
|
| 1944 |
+
"logits/chosen": -2.366685390472412,
|
| 1945 |
+
"logits/rejected": -2.244584798812866,
|
| 1946 |
+
"logps/chosen": -284.2767028808594,
|
| 1947 |
+
"logps/rejected": -525.98681640625,
|
| 1948 |
+
"loss": 0.1957,
|
| 1949 |
+
"positive_losses": 4.633613586425781,
|
| 1950 |
+
"rewards/accuracies": 1.0,
|
| 1951 |
+
"rewards/chosen": 0.3293871283531189,
|
| 1952 |
+
"rewards/margins": 3.1354575157165527,
|
| 1953 |
+
"rewards/margins_max": 4.131924629211426,
|
| 1954 |
+
"rewards/margins_min": 2.138990640640259,
|
| 1955 |
+
"rewards/margins_std": 1.409217119216919,
|
| 1956 |
+
"rewards/rejected": -2.806070327758789,
|
| 1957 |
+
"step": 880
|
| 1958 |
+
},
|
| 1959 |
+
{
|
| 1960 |
+
"dpo_losses": 0.07743240892887115,
|
| 1961 |
+
"epoch": 2.51,
|
| 1962 |
+
"grad_norm": 246.3549744525164,
|
| 1963 |
+
"learning_rate": 4.005001948670606e-07,
|
| 1964 |
+
"logits/chosen": -2.3962912559509277,
|
| 1965 |
+
"logits/rejected": -2.299287796020508,
|
| 1966 |
+
"logps/chosen": -425.3636169433594,
|
| 1967 |
+
"logps/rejected": -637.3418579101562,
|
| 1968 |
+
"loss": 0.2586,
|
| 1969 |
+
"positive_losses": 1.6243082284927368,
|
| 1970 |
+
"rewards/accuracies": 1.0,
|
| 1971 |
+
"rewards/chosen": 0.5158067941665649,
|
| 1972 |
+
"rewards/margins": 3.6285767555236816,
|
| 1973 |
+
"rewards/margins_max": 4.66379451751709,
|
| 1974 |
+
"rewards/margins_min": 2.593358278274536,
|
| 1975 |
+
"rewards/margins_std": 1.464019536972046,
|
| 1976 |
+
"rewards/rejected": -3.112769603729248,
|
| 1977 |
+
"step": 890
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"dpo_losses": 0.0913797914981842,
|
| 1981 |
+
"epoch": 2.54,
|
| 1982 |
+
"grad_norm": 31.034651555446352,
|
| 1983 |
+
"learning_rate": 3.571286656911377e-07,
|
| 1984 |
+
"logits/chosen": -2.356581211090088,
|
| 1985 |
+
"logits/rejected": -2.1674458980560303,
|
| 1986 |
+
"logps/chosen": -353.3656005859375,
|
| 1987 |
+
"logps/rejected": -581.6492309570312,
|
| 1988 |
+
"loss": 0.2348,
|
| 1989 |
+
"positive_losses": 2.3227431774139404,
|
| 1990 |
+
"rewards/accuracies": 1.0,
|
| 1991 |
+
"rewards/chosen": 0.46973830461502075,
|
| 1992 |
+
"rewards/margins": 3.632810115814209,
|
| 1993 |
+
"rewards/margins_max": 5.023132801055908,
|
| 1994 |
+
"rewards/margins_min": 2.242488145828247,
|
| 1995 |
+
"rewards/margins_std": 1.966212511062622,
|
| 1996 |
+
"rewards/rejected": -3.163072109222412,
|
| 1997 |
+
"step": 900
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 2.54,
|
| 2001 |
+
"eval_dpo_losses": 0.6903655529022217,
|
| 2002 |
+
"eval_logits/chosen": -2.3085381984710693,
|
| 2003 |
+
"eval_logits/rejected": -2.2660651206970215,
|
| 2004 |
+
"eval_logps/chosen": -430.1044006347656,
|
| 2005 |
+
"eval_logps/rejected": -459.71075439453125,
|
| 2006 |
+
"eval_loss": 14.72413444519043,
|
| 2007 |
+
"eval_positive_losses": 145.90809631347656,
|
| 2008 |
+
"eval_rewards/accuracies": 0.670634925365448,
|
| 2009 |
+
"eval_rewards/chosen": -1.4488320350646973,
|
| 2010 |
+
"eval_rewards/margins": 0.5564488768577576,
|
| 2011 |
+
"eval_rewards/margins_max": 2.180058240890503,
|
| 2012 |
+
"eval_rewards/margins_min": -1.0958278179168701,
|
| 2013 |
+
"eval_rewards/margins_std": 1.4586230516433716,
|
| 2014 |
+
"eval_rewards/rejected": -2.0052807331085205,
|
| 2015 |
+
"eval_runtime": 280.6688,
|
| 2016 |
+
"eval_samples_per_second": 7.126,
|
| 2017 |
+
"eval_steps_per_second": 0.224,
|
| 2018 |
+
"step": 900
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"dpo_losses": 0.10229980945587158,
|
| 2022 |
+
"epoch": 2.56,
|
| 2023 |
+
"grad_norm": 1.3207112449434741,
|
| 2024 |
+
"learning_rate": 3.1606136691612555e-07,
|
| 2025 |
+
"logits/chosen": -2.450596809387207,
|
| 2026 |
+
"logits/rejected": -2.320343255996704,
|
| 2027 |
+
"logps/chosen": -373.2110900878906,
|
| 2028 |
+
"logps/rejected": -512.347412109375,
|
| 2029 |
+
"loss": 0.2148,
|
| 2030 |
+
"positive_losses": 0.0,
|
| 2031 |
+
"rewards/accuracies": 1.0,
|
| 2032 |
+
"rewards/chosen": 0.45970940589904785,
|
| 2033 |
+
"rewards/margins": 3.199089527130127,
|
| 2034 |
+
"rewards/margins_max": 4.215831756591797,
|
| 2035 |
+
"rewards/margins_min": 2.182347297668457,
|
| 2036 |
+
"rewards/margins_std": 1.4378905296325684,
|
| 2037 |
+
"rewards/rejected": -2.739380121231079,
|
| 2038 |
+
"step": 910
|
| 2039 |
+
},
|
| 2040 |
+
{
|
| 2041 |
+
"dpo_losses": 0.10751441866159439,
|
| 2042 |
+
"epoch": 2.59,
|
| 2043 |
+
"grad_norm": 50.58623371052272,
|
| 2044 |
+
"learning_rate": 2.773424582247844e-07,
|
| 2045 |
+
"logits/chosen": -2.3134539127349854,
|
| 2046 |
+
"logits/rejected": -2.1395092010498047,
|
| 2047 |
+
"logps/chosen": -311.353271484375,
|
| 2048 |
+
"logps/rejected": -480.4183654785156,
|
| 2049 |
+
"loss": 0.1833,
|
| 2050 |
+
"positive_losses": 0.0,
|
| 2051 |
+
"rewards/accuracies": 1.0,
|
| 2052 |
+
"rewards/chosen": 0.4951690137386322,
|
| 2053 |
+
"rewards/margins": 3.3741965293884277,
|
| 2054 |
+
"rewards/margins_max": 4.4515910148620605,
|
| 2055 |
+
"rewards/margins_min": 2.296802043914795,
|
| 2056 |
+
"rewards/margins_std": 1.5236659049987793,
|
| 2057 |
+
"rewards/rejected": -2.8790273666381836,
|
| 2058 |
+
"step": 920
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"dpo_losses": 0.08198712766170502,
|
| 2062 |
+
"epoch": 2.62,
|
| 2063 |
+
"grad_norm": 34.59409580902137,
|
| 2064 |
+
"learning_rate": 2.410135740750821e-07,
|
| 2065 |
+
"logits/chosen": -2.3756449222564697,
|
| 2066 |
+
"logits/rejected": -2.2801589965820312,
|
| 2067 |
+
"logps/chosen": -309.6001892089844,
|
| 2068 |
+
"logps/rejected": -594.0984497070312,
|
| 2069 |
+
"loss": 0.2129,
|
| 2070 |
+
"positive_losses": 0.7936180233955383,
|
| 2071 |
+
"rewards/accuracies": 1.0,
|
| 2072 |
+
"rewards/chosen": 0.37005850672721863,
|
| 2073 |
+
"rewards/margins": 3.6145331859588623,
|
| 2074 |
+
"rewards/margins_max": 4.759096145629883,
|
| 2075 |
+
"rewards/margins_min": 2.4699695110321045,
|
| 2076 |
+
"rewards/margins_std": 1.618657112121582,
|
| 2077 |
+
"rewards/rejected": -3.244474411010742,
|
| 2078 |
+
"step": 930
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"dpo_losses": 0.18878915905952454,
|
| 2082 |
+
"epoch": 2.65,
|
| 2083 |
+
"grad_norm": 127.23596281859544,
|
| 2084 |
+
"learning_rate": 2.0711377893064182e-07,
|
| 2085 |
+
"logits/chosen": -2.362793207168579,
|
| 2086 |
+
"logits/rejected": -2.2321505546569824,
|
| 2087 |
+
"logps/chosen": -313.73736572265625,
|
| 2088 |
+
"logps/rejected": -428.6400451660156,
|
| 2089 |
+
"loss": 0.2036,
|
| 2090 |
+
"positive_losses": 2.974576711654663,
|
| 2091 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2092 |
+
"rewards/chosen": 0.2547670602798462,
|
| 2093 |
+
"rewards/margins": 2.439239978790283,
|
| 2094 |
+
"rewards/margins_max": 3.555748701095581,
|
| 2095 |
+
"rewards/margins_min": 1.3227306604385376,
|
| 2096 |
+
"rewards/margins_std": 1.5789823532104492,
|
| 2097 |
+
"rewards/rejected": -2.1844725608825684,
|
| 2098 |
+
"step": 940
|
| 2099 |
+
},
|
| 2100 |
+
{
|
| 2101 |
+
"dpo_losses": 0.1424863040447235,
|
| 2102 |
+
"epoch": 2.68,
|
| 2103 |
+
"grad_norm": 3.7675671139759395,
|
| 2104 |
+
"learning_rate": 1.756795252547111e-07,
|
| 2105 |
+
"logits/chosen": -2.324215888977051,
|
| 2106 |
+
"logits/rejected": -2.20219087600708,
|
| 2107 |
+
"logps/chosen": -260.1328430175781,
|
| 2108 |
+
"logps/rejected": -470.79412841796875,
|
| 2109 |
+
"loss": 0.3619,
|
| 2110 |
+
"positive_losses": 0.20402908325195312,
|
| 2111 |
+
"rewards/accuracies": 1.0,
|
| 2112 |
+
"rewards/chosen": 0.42012229561805725,
|
| 2113 |
+
"rewards/margins": 3.099902629852295,
|
| 2114 |
+
"rewards/margins_max": 4.072454452514648,
|
| 2115 |
+
"rewards/margins_min": 2.1273510456085205,
|
| 2116 |
+
"rewards/margins_std": 1.3753960132598877,
|
| 2117 |
+
"rewards/rejected": -2.6797804832458496,
|
| 2118 |
+
"step": 950
|
| 2119 |
+
},
|
| 2120 |
+
{
|
| 2121 |
+
"dpo_losses": 0.10174594074487686,
|
| 2122 |
+
"epoch": 2.7,
|
| 2123 |
+
"grad_norm": 1.8085505107547237,
|
| 2124 |
+
"learning_rate": 1.4674461431281013e-07,
|
| 2125 |
+
"logits/chosen": -2.525442361831665,
|
| 2126 |
+
"logits/rejected": -2.4085853099823,
|
| 2127 |
+
"logps/chosen": -290.04345703125,
|
| 2128 |
+
"logps/rejected": -531.1943969726562,
|
| 2129 |
+
"loss": 0.3038,
|
| 2130 |
+
"positive_losses": 0.1788475066423416,
|
| 2131 |
+
"rewards/accuracies": 1.0,
|
| 2132 |
+
"rewards/chosen": 0.3685051202774048,
|
| 2133 |
+
"rewards/margins": 3.2411983013153076,
|
| 2134 |
+
"rewards/margins_max": 4.235571384429932,
|
| 2135 |
+
"rewards/margins_min": 2.246825695037842,
|
| 2136 |
+
"rewards/margins_std": 1.4062554836273193,
|
| 2137 |
+
"rewards/rejected": -2.8726933002471924,
|
| 2138 |
+
"step": 960
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"dpo_losses": 0.09436773508787155,
|
| 2142 |
+
"epoch": 2.73,
|
| 2143 |
+
"grad_norm": 38.21579384775439,
|
| 2144 |
+
"learning_rate": 1.2034015982622243e-07,
|
| 2145 |
+
"logits/chosen": -2.403714418411255,
|
| 2146 |
+
"logits/rejected": -2.282444477081299,
|
| 2147 |
+
"logps/chosen": -336.4191589355469,
|
| 2148 |
+
"logps/rejected": -650.2848510742188,
|
| 2149 |
+
"loss": 0.2727,
|
| 2150 |
+
"positive_losses": 4.7762908935546875,
|
| 2151 |
+
"rewards/accuracies": 1.0,
|
| 2152 |
+
"rewards/chosen": 0.38717547059059143,
|
| 2153 |
+
"rewards/margins": 3.8552498817443848,
|
| 2154 |
+
"rewards/margins_max": 5.095088481903076,
|
| 2155 |
+
"rewards/margins_min": 2.615410327911377,
|
| 2156 |
+
"rewards/margins_std": 1.7533977031707764,
|
| 2157 |
+
"rewards/rejected": -3.468074083328247,
|
| 2158 |
+
"step": 970
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"dpo_losses": 0.0995684489607811,
|
| 2162 |
+
"epoch": 2.76,
|
| 2163 |
+
"grad_norm": 34.214640164153415,
|
| 2164 |
+
"learning_rate": 9.649455451539419e-08,
|
| 2165 |
+
"logits/chosen": -2.195786714553833,
|
| 2166 |
+
"logits/rejected": -2.1557459831237793,
|
| 2167 |
+
"logps/chosen": -205.28829956054688,
|
| 2168 |
+
"logps/rejected": -449.69903564453125,
|
| 2169 |
+
"loss": 0.362,
|
| 2170 |
+
"positive_losses": 1.5731815099716187,
|
| 2171 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2172 |
+
"rewards/chosen": 0.3235073983669281,
|
| 2173 |
+
"rewards/margins": 3.076923131942749,
|
| 2174 |
+
"rewards/margins_max": 3.9705424308776855,
|
| 2175 |
+
"rewards/margins_min": 2.1833040714263916,
|
| 2176 |
+
"rewards/margins_std": 1.2637684345245361,
|
| 2177 |
+
"rewards/rejected": -2.753415584564209,
|
| 2178 |
+
"step": 980
|
| 2179 |
+
},
|
| 2180 |
+
{
|
| 2181 |
+
"dpo_losses": 0.09841219335794449,
|
| 2182 |
+
"epoch": 2.79,
|
| 2183 |
+
"grad_norm": 3.551041227418815,
|
| 2184 |
+
"learning_rate": 7.523343956923196e-08,
|
| 2185 |
+
"logits/chosen": -2.4305670261383057,
|
| 2186 |
+
"logits/rejected": -2.3597071170806885,
|
| 2187 |
+
"logps/chosen": -302.9432678222656,
|
| 2188 |
+
"logps/rejected": -600.987060546875,
|
| 2189 |
+
"loss": 0.2684,
|
| 2190 |
+
"positive_losses": 1.013157606124878,
|
| 2191 |
+
"rewards/accuracies": 1.0,
|
| 2192 |
+
"rewards/chosen": 0.49134930968284607,
|
| 2193 |
+
"rewards/margins": 3.6477818489074707,
|
| 2194 |
+
"rewards/margins_max": 4.790149688720703,
|
| 2195 |
+
"rewards/margins_min": 2.5054140090942383,
|
| 2196 |
+
"rewards/margins_std": 1.6155517101287842,
|
| 2197 |
+
"rewards/rejected": -3.156432628631592,
|
| 2198 |
+
"step": 990
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"dpo_losses": 0.11016283929347992,
|
| 2202 |
+
"epoch": 2.82,
|
| 2203 |
+
"grad_norm": 17.885804652199482,
|
| 2204 |
+
"learning_rate": 5.657967707312195e-08,
|
| 2205 |
+
"logits/chosen": -2.2564501762390137,
|
| 2206 |
+
"logits/rejected": -2.223057985305786,
|
| 2207 |
+
"logps/chosen": -243.24911499023438,
|
| 2208 |
+
"logps/rejected": -581.9591064453125,
|
| 2209 |
+
"loss": 0.1369,
|
| 2210 |
+
"positive_losses": 0.5390418171882629,
|
| 2211 |
+
"rewards/accuracies": 1.0,
|
| 2212 |
+
"rewards/chosen": 0.3867679238319397,
|
| 2213 |
+
"rewards/margins": 3.3442111015319824,
|
| 2214 |
+
"rewards/margins_max": 4.438827991485596,
|
| 2215 |
+
"rewards/margins_min": 2.2495944499969482,
|
| 2216 |
+
"rewards/margins_std": 1.5480217933654785,
|
| 2217 |
+
"rewards/rejected": -2.9574437141418457,
|
| 2218 |
+
"step": 1000
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"epoch": 2.82,
|
| 2222 |
+
"eval_dpo_losses": 0.6868842244148254,
|
| 2223 |
+
"eval_logits/chosen": -2.3165292739868164,
|
| 2224 |
+
"eval_logits/rejected": -2.2738356590270996,
|
| 2225 |
+
"eval_logps/chosen": -428.1341857910156,
|
| 2226 |
+
"eval_logps/rejected": -461.68865966796875,
|
| 2227 |
+
"eval_loss": 14.59554386138916,
|
| 2228 |
+
"eval_positive_losses": 143.93893432617188,
|
| 2229 |
+
"eval_rewards/accuracies": 0.6626983880996704,
|
| 2230 |
+
"eval_rewards/chosen": -1.4291293621063232,
|
| 2231 |
+
"eval_rewards/margins": 0.5959304571151733,
|
| 2232 |
+
"eval_rewards/margins_max": 2.2952685356140137,
|
| 2233 |
+
"eval_rewards/margins_min": -1.107314109802246,
|
| 2234 |
+
"eval_rewards/margins_std": 1.5052008628845215,
|
| 2235 |
+
"eval_rewards/rejected": -2.025059700012207,
|
| 2236 |
+
"eval_runtime": 280.8142,
|
| 2237 |
+
"eval_samples_per_second": 7.122,
|
| 2238 |
+
"eval_steps_per_second": 0.224,
|
| 2239 |
+
"step": 1000
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"dpo_losses": 0.13428188860416412,
|
| 2243 |
+
"epoch": 2.85,
|
| 2244 |
+
"grad_norm": 223.7670035438648,
|
| 2245 |
+
"learning_rate": 4.055332542531959e-08,
|
| 2246 |
+
"logits/chosen": -2.418339490890503,
|
| 2247 |
+
"logits/rejected": -2.3237392902374268,
|
| 2248 |
+
"logps/chosen": -274.25958251953125,
|
| 2249 |
+
"logps/rejected": -552.5701904296875,
|
| 2250 |
+
"loss": 0.4402,
|
| 2251 |
+
"positive_losses": 0.35679930448532104,
|
| 2252 |
+
"rewards/accuracies": 1.0,
|
| 2253 |
+
"rewards/chosen": 0.41407138109207153,
|
| 2254 |
+
"rewards/margins": 3.3448989391326904,
|
| 2255 |
+
"rewards/margins_max": 4.5948896408081055,
|
| 2256 |
+
"rewards/margins_min": 2.094907522201538,
|
| 2257 |
+
"rewards/margins_std": 1.7677549123764038,
|
| 2258 |
+
"rewards/rejected": -2.9308273792266846,
|
| 2259 |
+
"step": 1010
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"dpo_losses": 0.09611718356609344,
|
| 2263 |
+
"epoch": 2.87,
|
| 2264 |
+
"grad_norm": 4.2164881541545896,
|
| 2265 |
+
"learning_rate": 2.7171617768147472e-08,
|
| 2266 |
+
"logits/chosen": -2.3242077827453613,
|
| 2267 |
+
"logits/rejected": -2.20931339263916,
|
| 2268 |
+
"logps/chosen": -273.04669189453125,
|
| 2269 |
+
"logps/rejected": -532.8912963867188,
|
| 2270 |
+
"loss": 0.2835,
|
| 2271 |
+
"positive_losses": 1.0188411474227905,
|
| 2272 |
+
"rewards/accuracies": 1.0,
|
| 2273 |
+
"rewards/chosen": 0.38434693217277527,
|
| 2274 |
+
"rewards/margins": 3.382856845855713,
|
| 2275 |
+
"rewards/margins_max": 4.1625471115112305,
|
| 2276 |
+
"rewards/margins_min": 2.6031665802001953,
|
| 2277 |
+
"rewards/margins_std": 1.1026487350463867,
|
| 2278 |
+
"rewards/rejected": -2.9985098838806152,
|
| 2279 |
+
"step": 1020
|
| 2280 |
+
},
|
| 2281 |
+
{
|
| 2282 |
+
"dpo_losses": 0.09702328592538834,
|
| 2283 |
+
"epoch": 2.9,
|
| 2284 |
+
"grad_norm": 37.444607526485036,
|
| 2285 |
+
"learning_rate": 1.6448943457189616e-08,
|
| 2286 |
+
"logits/chosen": -2.3770041465759277,
|
| 2287 |
+
"logits/rejected": -2.312561511993408,
|
| 2288 |
+
"logps/chosen": -323.7328186035156,
|
| 2289 |
+
"logps/rejected": -590.1715698242188,
|
| 2290 |
+
"loss": 0.1692,
|
| 2291 |
+
"positive_losses": 0.22979411482810974,
|
| 2292 |
+
"rewards/accuracies": 1.0,
|
| 2293 |
+
"rewards/chosen": 0.4143516421318054,
|
| 2294 |
+
"rewards/margins": 3.550870418548584,
|
| 2295 |
+
"rewards/margins_max": 4.831820487976074,
|
| 2296 |
+
"rewards/margins_min": 2.269920825958252,
|
| 2297 |
+
"rewards/margins_std": 1.8115367889404297,
|
| 2298 |
+
"rewards/rejected": -3.136518955230713,
|
| 2299 |
+
"step": 1030
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"dpo_losses": 0.05908944085240364,
|
| 2303 |
+
"epoch": 2.93,
|
| 2304 |
+
"grad_norm": 4.414870466706013,
|
| 2305 |
+
"learning_rate": 8.39683258841123e-09,
|
| 2306 |
+
"logits/chosen": -2.2308764457702637,
|
| 2307 |
+
"logits/rejected": -2.1003758907318115,
|
| 2308 |
+
"logps/chosen": -295.9768371582031,
|
| 2309 |
+
"logps/rejected": -529.7901611328125,
|
| 2310 |
+
"loss": 0.2377,
|
| 2311 |
+
"positive_losses": 3.635768175125122,
|
| 2312 |
+
"rewards/accuracies": 1.0,
|
| 2313 |
+
"rewards/chosen": 0.5024327039718628,
|
| 2314 |
+
"rewards/margins": 3.5135796070098877,
|
| 2315 |
+
"rewards/margins_max": 4.2515668869018555,
|
| 2316 |
+
"rewards/margins_min": 2.7755913734436035,
|
| 2317 |
+
"rewards/margins_std": 1.0436723232269287,
|
| 2318 |
+
"rewards/rejected": -3.0111465454101562,
|
| 2319 |
+
"step": 1040
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"dpo_losses": 0.13825824856758118,
|
| 2323 |
+
"epoch": 2.96,
|
| 2324 |
+
"grad_norm": 2.4504883804901993,
|
| 2325 |
+
"learning_rate": 3.0239435998430376e-09,
|
| 2326 |
+
"logits/chosen": -2.348167896270752,
|
| 2327 |
+
"logits/rejected": -2.233564615249634,
|
| 2328 |
+
"logps/chosen": -286.636962890625,
|
| 2329 |
+
"logps/rejected": -520.432861328125,
|
| 2330 |
+
"loss": 0.2494,
|
| 2331 |
+
"positive_losses": 3.874138593673706,
|
| 2332 |
+
"rewards/accuracies": 1.0,
|
| 2333 |
+
"rewards/chosen": 0.4494766294956207,
|
| 2334 |
+
"rewards/margins": 3.075880527496338,
|
| 2335 |
+
"rewards/margins_max": 4.244940280914307,
|
| 2336 |
+
"rewards/margins_min": 1.9068210124969482,
|
| 2337 |
+
"rewards/margins_std": 1.6533000469207764,
|
| 2338 |
+
"rewards/rejected": -2.62640380859375,
|
| 2339 |
+
"step": 1050
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"dpo_losses": 0.10627589374780655,
|
| 2343 |
+
"epoch": 2.99,
|
| 2344 |
+
"grad_norm": 2.8525382266880834,
|
| 2345 |
+
"learning_rate": 3.3605396115826695e-10,
|
| 2346 |
+
"logits/chosen": -2.2836787700653076,
|
| 2347 |
+
"logits/rejected": -2.183605432510376,
|
| 2348 |
+
"logps/chosen": -289.5533447265625,
|
| 2349 |
+
"logps/rejected": -472.84375,
|
| 2350 |
+
"loss": 0.3088,
|
| 2351 |
+
"positive_losses": 0.0,
|
| 2352 |
+
"rewards/accuracies": 1.0,
|
| 2353 |
+
"rewards/chosen": 0.3630369305610657,
|
| 2354 |
+
"rewards/margins": 2.99712872505188,
|
| 2355 |
+
"rewards/margins_max": 3.9373364448547363,
|
| 2356 |
+
"rewards/margins_min": 2.0569205284118652,
|
| 2357 |
+
"rewards/margins_std": 1.3296549320220947,
|
| 2358 |
+
"rewards/rejected": -2.63409161567688,
|
| 2359 |
+
"step": 1060
|
| 2360 |
+
},
|
| 2361 |
+
{
|
| 2362 |
+
"epoch": 3.0,
|
| 2363 |
+
"step": 1065,
|
| 2364 |
+
"total_flos": 0.0,
|
| 2365 |
+
"train_loss": 0.36563416943303856,
|
| 2366 |
+
"train_runtime": 9271.2095,
|
| 2367 |
+
"train_samples_per_second": 1.837,
|
| 2368 |
+
"train_steps_per_second": 0.115
|
| 2369 |
+
}
|
| 2370 |
+
],
|
| 2371 |
+
"logging_steps": 10,
|
| 2372 |
+
"max_steps": 1065,
|
| 2373 |
+
"num_input_tokens_seen": 0,
|
| 2374 |
+
"num_train_epochs": 3,
|
| 2375 |
+
"save_steps": 100,
|
| 2376 |
+
"total_flos": 0.0,
|
| 2377 |
+
"train_batch_size": 2,
|
| 2378 |
+
"trial_name": null,
|
| 2379 |
+
"trial_params": null
|
| 2380 |
+
}
|