Model save
Browse files- README.md +79 -0
- adapter_config.json +3 -3
- adapter_model.safetensors +1 -1
- all_results.json +21 -0
- eval_results.json +16 -0
- runs/Mar10_08-34-52_66ae9c49b9d0/events.out.tfevents.1710059740.66ae9c49b9d0.36901.0 +2 -2
- runs/Mar10_11-42-58_38f918e859f8/events.out.tfevents.1710071153.38f918e859f8.7381.0 +3 -0
- runs/Mar10_11-42-58_38f918e859f8/events.out.tfevents.1710072347.38f918e859f8.7381.1 +3 -0
- train_results.json +8 -0
- trainer_state.json +1418 -0
- training_args.bin +1 -1
README.md
ADDED
|
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: peft
|
| 3 |
+
tags:
|
| 4 |
+
- trl
|
| 5 |
+
- dpo
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
base_model: meta-llama/Llama-2-7b-chat-hf
|
| 8 |
+
model-index:
|
| 9 |
+
- name: cira-7b-dpo-lora
|
| 10 |
+
results: []
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 14 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 15 |
+
|
| 16 |
+
# cira-7b-dpo-lora
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [meta-llama/Llama-2-7b-chat-hf](https://huggingface.co/meta-llama/Llama-2-7b-chat-hf) on the None dataset.
|
| 19 |
+
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.6183
|
| 21 |
+
- Rewards/chosen: 0.5535
|
| 22 |
+
- Rewards/rejected: 0.3385
|
| 23 |
+
- Rewards/accuracies: 0.6784
|
| 24 |
+
- Rewards/margins: 0.2150
|
| 25 |
+
- Logps/rejected: -652.2422
|
| 26 |
+
- Logps/chosen: -795.1126
|
| 27 |
+
- Logits/rejected: -1.1812
|
| 28 |
+
- Logits/chosen: -1.0305
|
| 29 |
+
|
| 30 |
+
## Model description
|
| 31 |
+
|
| 32 |
+
More information needed
|
| 33 |
+
|
| 34 |
+
## Intended uses & limitations
|
| 35 |
+
|
| 36 |
+
More information needed
|
| 37 |
+
|
| 38 |
+
## Training and evaluation data
|
| 39 |
+
|
| 40 |
+
More information needed
|
| 41 |
+
|
| 42 |
+
## Training procedure
|
| 43 |
+
|
| 44 |
+
### Training hyperparameters
|
| 45 |
+
|
| 46 |
+
The following hyperparameters were used during training:
|
| 47 |
+
- learning_rate: 5e-06
|
| 48 |
+
- train_batch_size: 1
|
| 49 |
+
- eval_batch_size: 1
|
| 50 |
+
- seed: 42
|
| 51 |
+
- distributed_type: multi-GPU
|
| 52 |
+
- gradient_accumulation_steps: 4
|
| 53 |
+
- total_train_batch_size: 4
|
| 54 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 55 |
+
- lr_scheduler_type: cosine
|
| 56 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 57 |
+
- num_epochs: 1
|
| 58 |
+
|
| 59 |
+
### Training results
|
| 60 |
+
|
| 61 |
+
| Training Loss | Epoch | Step | Logits/chosen | Logits/rejected | Logps/chosen | Logps/rejected | Validation Loss | Rewards/accuracies | Rewards/chosen | Rewards/margins | Rewards/rejected |
|
| 62 |
+
|:-------------:|:-----:|:----:|:-------------:|:---------------:|:------------:|:--------------:|:---------------:|:------------------:|:--------------:|:---------------:|:----------------:|
|
| 63 |
+
| 0.6618 | 0.11 | 100 | -0.8082 | -1.0029 | -823.6102 | -665.3923 | 0.6664 | 0.6432 | 0.2685 | 0.0615 | 0.2070 |
|
| 64 |
+
| 0.6079 | 0.22 | 200 | -1.0530 | -1.2188 | -794.3279 | -642.6389 | 0.6463 | 0.6508 | 0.5613 | 0.1268 | 0.4345 |
|
| 65 |
+
| 0.6029 | 0.33 | 300 | -1.0367 | -1.1965 | -793.2078 | -644.8513 | 0.6360 | 0.6558 | 0.5725 | 0.1601 | 0.4124 |
|
| 66 |
+
| 0.6123 | 0.45 | 400 | -1.1220 | -1.2658 | -787.7750 | -641.9633 | 0.6291 | 0.6608 | 0.6269 | 0.1856 | 0.4413 |
|
| 67 |
+
| 0.5596 | 0.56 | 500 | -1.0852 | -1.2330 | -790.7928 | -646.7930 | 0.6230 | 0.6683 | 0.5967 | 0.2037 | 0.3930 |
|
| 68 |
+
| 0.5382 | 0.67 | 600 | -1.0547 | -1.2034 | -793.2486 | -650.0926 | 0.6199 | 0.6709 | 0.5721 | 0.2121 | 0.3600 |
|
| 69 |
+
| 0.5952 | 0.78 | 700 | -1.0324 | -1.1827 | -794.9604 | -652.0420 | 0.6186 | 0.6784 | 0.5550 | 0.2145 | 0.3405 |
|
| 70 |
+
| 0.5792 | 0.89 | 800 | 0.6182 | 0.5534 | 0.3382 | 0.6784 | 0.2151 | -652.2705 | -795.125 | -1.1812 | -1.0308 |
|
| 71 |
+
|
| 72 |
+
|
| 73 |
+
### Framework versions
|
| 74 |
+
|
| 75 |
+
- PEFT 0.9.0
|
| 76 |
+
- Transformers 4.36.2
|
| 77 |
+
- Pytorch 2.1.0+cu121
|
| 78 |
+
- Datasets 2.14.6
|
| 79 |
+
- Tokenizers 0.15.2
|
adapter_config.json
CHANGED
|
@@ -19,12 +19,12 @@
|
|
| 19 |
"rank_pattern": {},
|
| 20 |
"revision": null,
|
| 21 |
"target_modules": [
|
|
|
|
|
|
|
| 22 |
"q_proj",
|
|
|
|
| 23 |
"v_proj",
|
| 24 |
"gate_proj",
|
| 25 |
-
"k_proj",
|
| 26 |
-
"up_proj",
|
| 27 |
-
"o_proj",
|
| 28 |
"down_proj"
|
| 29 |
],
|
| 30 |
"task_type": "CAUSAL_LM",
|
|
|
|
| 19 |
"rank_pattern": {},
|
| 20 |
"revision": null,
|
| 21 |
"target_modules": [
|
| 22 |
+
"o_proj",
|
| 23 |
+
"k_proj",
|
| 24 |
"q_proj",
|
| 25 |
+
"up_proj",
|
| 26 |
"v_proj",
|
| 27 |
"gate_proj",
|
|
|
|
|
|
|
|
|
|
| 28 |
"down_proj"
|
| 29 |
],
|
| 30 |
"task_type": "CAUSAL_LM",
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 639692768
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2e8c05c1be22d860591910dc0a0064aa624f6e08eac1d7faec0cf6008174ff3a
|
| 3 |
size 639692768
|
all_results.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_logits/chosen": -1.030522108078003,
|
| 4 |
+
"eval_logits/rejected": -1.1812418699264526,
|
| 5 |
+
"eval_logps/chosen": -795.1126098632812,
|
| 6 |
+
"eval_logps/rejected": -652.2422485351562,
|
| 7 |
+
"eval_loss": 0.6183284521102905,
|
| 8 |
+
"eval_rewards/accuracies": 0.6783919334411621,
|
| 9 |
+
"eval_rewards/chosen": 0.5534913539886475,
|
| 10 |
+
"eval_rewards/margins": 0.214975506067276,
|
| 11 |
+
"eval_rewards/rejected": 0.33851587772369385,
|
| 12 |
+
"eval_runtime": 182.1641,
|
| 13 |
+
"eval_samples": 398,
|
| 14 |
+
"eval_samples_per_second": 2.185,
|
| 15 |
+
"eval_steps_per_second": 2.185,
|
| 16 |
+
"train_loss": 0.12323601163483516,
|
| 17 |
+
"train_runtime": 1012.2753,
|
| 18 |
+
"train_samples": 3588,
|
| 19 |
+
"train_samples_per_second": 3.544,
|
| 20 |
+
"train_steps_per_second": 0.886
|
| 21 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_logits/chosen": -1.030522108078003,
|
| 4 |
+
"eval_logits/rejected": -1.1812418699264526,
|
| 5 |
+
"eval_logps/chosen": -795.1126098632812,
|
| 6 |
+
"eval_logps/rejected": -652.2422485351562,
|
| 7 |
+
"eval_loss": 0.6183284521102905,
|
| 8 |
+
"eval_rewards/accuracies": 0.6783919334411621,
|
| 9 |
+
"eval_rewards/chosen": 0.5534913539886475,
|
| 10 |
+
"eval_rewards/margins": 0.214975506067276,
|
| 11 |
+
"eval_rewards/rejected": 0.33851587772369385,
|
| 12 |
+
"eval_runtime": 182.1641,
|
| 13 |
+
"eval_samples": 398,
|
| 14 |
+
"eval_samples_per_second": 2.185,
|
| 15 |
+
"eval_steps_per_second": 2.185
|
| 16 |
+
}
|
runs/Mar10_08-34-52_66ae9c49b9d0/events.out.tfevents.1710059740.66ae9c49b9d0.36901.0
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:e06fe475aac9525914ab9224a5895997a84163d2f872a997e33a88448693f871
|
| 3 |
+
size 60705
|
runs/Mar10_11-42-58_38f918e859f8/events.out.tfevents.1710071153.38f918e859f8.7381.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f4bc5e7f9ff43cf611de1558bdbdb1cd9ae38c9da0a7763d7436fa54168f2340
|
| 3 |
+
size 18101
|
runs/Mar10_11-42-58_38f918e859f8/events.out.tfevents.1710072347.38f918e859f8.7381.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bd5190bc9be8757a35ff0b005c44da4826c935848aa2a3ef32ca698fc6292402
|
| 3 |
+
size 828
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.12323601163483516,
|
| 4 |
+
"train_runtime": 1012.2753,
|
| 5 |
+
"train_samples": 3588,
|
| 6 |
+
"train_samples_per_second": 3.544,
|
| 7 |
+
"train_steps_per_second": 0.886
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1418 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 897,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 5.555555555555556e-08,
|
| 14 |
+
"logits/chosen": -0.5698858499526978,
|
| 15 |
+
"logits/rejected": -0.7608428597450256,
|
| 16 |
+
"logps/chosen": -628.0617065429688,
|
| 17 |
+
"logps/rejected": -619.150390625,
|
| 18 |
+
"loss": 0.6931,
|
| 19 |
+
"rewards/accuracies": 0.0,
|
| 20 |
+
"rewards/chosen": 0.0,
|
| 21 |
+
"rewards/margins": 0.0,
|
| 22 |
+
"rewards/rejected": 0.0,
|
| 23 |
+
"step": 1
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.01,
|
| 27 |
+
"learning_rate": 5.555555555555555e-07,
|
| 28 |
+
"logits/chosen": -0.6823484897613525,
|
| 29 |
+
"logits/rejected": -0.8546125888824463,
|
| 30 |
+
"logps/chosen": -777.4924926757812,
|
| 31 |
+
"logps/rejected": -634.3888549804688,
|
| 32 |
+
"loss": 0.6933,
|
| 33 |
+
"rewards/accuracies": 0.3611111044883728,
|
| 34 |
+
"rewards/chosen": -0.0005514526856131852,
|
| 35 |
+
"rewards/margins": -0.00033656222512945533,
|
| 36 |
+
"rewards/rejected": -0.00021489031496457756,
|
| 37 |
+
"step": 10
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.02,
|
| 41 |
+
"learning_rate": 1.111111111111111e-06,
|
| 42 |
+
"logits/chosen": -0.7127803564071655,
|
| 43 |
+
"logits/rejected": -0.8547202348709106,
|
| 44 |
+
"logps/chosen": -899.9298095703125,
|
| 45 |
+
"logps/rejected": -712.7205810546875,
|
| 46 |
+
"loss": 0.6912,
|
| 47 |
+
"rewards/accuracies": 0.550000011920929,
|
| 48 |
+
"rewards/chosen": 0.005615489557385445,
|
| 49 |
+
"rewards/margins": 0.003956252709031105,
|
| 50 |
+
"rewards/rejected": 0.0016592368483543396,
|
| 51 |
+
"step": 20
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.03,
|
| 55 |
+
"learning_rate": 1.6666666666666667e-06,
|
| 56 |
+
"logits/chosen": -0.702410101890564,
|
| 57 |
+
"logits/rejected": -0.9075536727905273,
|
| 58 |
+
"logps/chosen": -799.2589111328125,
|
| 59 |
+
"logps/rejected": -717.34716796875,
|
| 60 |
+
"loss": 0.6903,
|
| 61 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 62 |
+
"rewards/chosen": 0.01230540033429861,
|
| 63 |
+
"rewards/margins": 0.005792214069515467,
|
| 64 |
+
"rewards/rejected": 0.006513187196105719,
|
| 65 |
+
"step": 30
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.04,
|
| 69 |
+
"learning_rate": 2.222222222222222e-06,
|
| 70 |
+
"logits/chosen": -0.7552632093429565,
|
| 71 |
+
"logits/rejected": -0.9524158239364624,
|
| 72 |
+
"logps/chosen": -835.037109375,
|
| 73 |
+
"logps/rejected": -631.7636108398438,
|
| 74 |
+
"loss": 0.6904,
|
| 75 |
+
"rewards/accuracies": 0.574999988079071,
|
| 76 |
+
"rewards/chosen": 0.025444606319069862,
|
| 77 |
+
"rewards/margins": 0.005691047292202711,
|
| 78 |
+
"rewards/rejected": 0.019753558561205864,
|
| 79 |
+
"step": 40
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.06,
|
| 83 |
+
"learning_rate": 2.7777777777777783e-06,
|
| 84 |
+
"logits/chosen": -0.7381051778793335,
|
| 85 |
+
"logits/rejected": -0.8419072031974792,
|
| 86 |
+
"logps/chosen": -788.6356811523438,
|
| 87 |
+
"logps/rejected": -730.46728515625,
|
| 88 |
+
"loss": 0.6901,
|
| 89 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 90 |
+
"rewards/chosen": 0.04884253069758415,
|
| 91 |
+
"rewards/margins": 0.006506313569843769,
|
| 92 |
+
"rewards/rejected": 0.04233621805906296,
|
| 93 |
+
"step": 50
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.07,
|
| 97 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 98 |
+
"logits/chosen": -0.6829635500907898,
|
| 99 |
+
"logits/rejected": -0.8546798825263977,
|
| 100 |
+
"logps/chosen": -803.8533325195312,
|
| 101 |
+
"logps/rejected": -685.8370361328125,
|
| 102 |
+
"loss": 0.684,
|
| 103 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 104 |
+
"rewards/chosen": 0.07093875110149384,
|
| 105 |
+
"rewards/margins": 0.01889834739267826,
|
| 106 |
+
"rewards/rejected": 0.052040405571460724,
|
| 107 |
+
"step": 60
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.08,
|
| 111 |
+
"learning_rate": 3.88888888888889e-06,
|
| 112 |
+
"logits/chosen": -0.6931325197219849,
|
| 113 |
+
"logits/rejected": -0.8457029461860657,
|
| 114 |
+
"logps/chosen": -835.8101806640625,
|
| 115 |
+
"logps/rejected": -559.0534057617188,
|
| 116 |
+
"loss": 0.674,
|
| 117 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 118 |
+
"rewards/chosen": 0.11979933828115463,
|
| 119 |
+
"rewards/margins": 0.04078071564435959,
|
| 120 |
+
"rewards/rejected": 0.07901863008737564,
|
| 121 |
+
"step": 70
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.09,
|
| 125 |
+
"learning_rate": 4.444444444444444e-06,
|
| 126 |
+
"logits/chosen": -0.7829106450080872,
|
| 127 |
+
"logits/rejected": -1.0052988529205322,
|
| 128 |
+
"logps/chosen": -857.0613403320312,
|
| 129 |
+
"logps/rejected": -581.3321533203125,
|
| 130 |
+
"loss": 0.6535,
|
| 131 |
+
"rewards/accuracies": 0.625,
|
| 132 |
+
"rewards/chosen": 0.18766728043556213,
|
| 133 |
+
"rewards/margins": 0.08457548916339874,
|
| 134 |
+
"rewards/rejected": 0.10309179872274399,
|
| 135 |
+
"step": 80
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.1,
|
| 139 |
+
"learning_rate": 5e-06,
|
| 140 |
+
"logits/chosen": -0.8815785646438599,
|
| 141 |
+
"logits/rejected": -0.96449214220047,
|
| 142 |
+
"logps/chosen": -789.5969848632812,
|
| 143 |
+
"logps/rejected": -702.9019165039062,
|
| 144 |
+
"loss": 0.6794,
|
| 145 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 146 |
+
"rewards/chosen": 0.20566365122795105,
|
| 147 |
+
"rewards/margins": 0.033137936145067215,
|
| 148 |
+
"rewards/rejected": 0.17252573370933533,
|
| 149 |
+
"step": 90
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.11,
|
| 153 |
+
"learning_rate": 4.9981058784687895e-06,
|
| 154 |
+
"logits/chosen": -0.8663153648376465,
|
| 155 |
+
"logits/rejected": -1.0621612071990967,
|
| 156 |
+
"logps/chosen": -868.1552734375,
|
| 157 |
+
"logps/rejected": -664.0401611328125,
|
| 158 |
+
"loss": 0.6618,
|
| 159 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 160 |
+
"rewards/chosen": 0.2395782470703125,
|
| 161 |
+
"rewards/margins": 0.06794992089271545,
|
| 162 |
+
"rewards/rejected": 0.17162834107875824,
|
| 163 |
+
"step": 100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.11,
|
| 167 |
+
"eval_logits/chosen": -0.8082190752029419,
|
| 168 |
+
"eval_logits/rejected": -1.0028988122940063,
|
| 169 |
+
"eval_logps/chosen": -823.6102294921875,
|
| 170 |
+
"eval_logps/rejected": -665.3922729492188,
|
| 171 |
+
"eval_loss": 0.6664446592330933,
|
| 172 |
+
"eval_rewards/accuracies": 0.643216073513031,
|
| 173 |
+
"eval_rewards/chosen": 0.268514484167099,
|
| 174 |
+
"eval_rewards/margins": 0.061498455703258514,
|
| 175 |
+
"eval_rewards/rejected": 0.20701603591442108,
|
| 176 |
+
"eval_runtime": 181.7312,
|
| 177 |
+
"eval_samples_per_second": 2.19,
|
| 178 |
+
"eval_steps_per_second": 2.19,
|
| 179 |
+
"step": 100
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.12,
|
| 183 |
+
"learning_rate": 4.992426384032258e-06,
|
| 184 |
+
"logits/chosen": -0.8113915324211121,
|
| 185 |
+
"logits/rejected": -1.1199345588684082,
|
| 186 |
+
"logps/chosen": -766.8582763671875,
|
| 187 |
+
"logps/rejected": -536.8991088867188,
|
| 188 |
+
"loss": 0.6432,
|
| 189 |
+
"rewards/accuracies": 0.824999988079071,
|
| 190 |
+
"rewards/chosen": 0.2860961854457855,
|
| 191 |
+
"rewards/margins": 0.11079935729503632,
|
| 192 |
+
"rewards/rejected": 0.1752968281507492,
|
| 193 |
+
"step": 110
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.13,
|
| 197 |
+
"learning_rate": 4.982970122812566e-06,
|
| 198 |
+
"logits/chosen": -0.8754690289497375,
|
| 199 |
+
"logits/rejected": -0.9867092370986938,
|
| 200 |
+
"logps/chosen": -783.0337524414062,
|
| 201 |
+
"logps/rejected": -615.1319580078125,
|
| 202 |
+
"loss": 0.6622,
|
| 203 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 204 |
+
"rewards/chosen": 0.3586369454860687,
|
| 205 |
+
"rewards/margins": 0.08097393810749054,
|
| 206 |
+
"rewards/rejected": 0.277662992477417,
|
| 207 |
+
"step": 120
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.14,
|
| 211 |
+
"learning_rate": 4.969751423856095e-06,
|
| 212 |
+
"logits/chosen": -0.8801660537719727,
|
| 213 |
+
"logits/rejected": -1.0285800695419312,
|
| 214 |
+
"logps/chosen": -857.1593017578125,
|
| 215 |
+
"logps/rejected": -575.841796875,
|
| 216 |
+
"loss": 0.6472,
|
| 217 |
+
"rewards/accuracies": 0.625,
|
| 218 |
+
"rewards/chosen": 0.41047531366348267,
|
| 219 |
+
"rewards/margins": 0.11212178319692612,
|
| 220 |
+
"rewards/rejected": 0.29835352301597595,
|
| 221 |
+
"step": 130
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.16,
|
| 225 |
+
"learning_rate": 4.952790317420694e-06,
|
| 226 |
+
"logits/chosen": -0.9968269467353821,
|
| 227 |
+
"logits/rejected": -1.1242711544036865,
|
| 228 |
+
"logps/chosen": -922.9993286132812,
|
| 229 |
+
"logps/rejected": -759.6329345703125,
|
| 230 |
+
"loss": 0.6605,
|
| 231 |
+
"rewards/accuracies": 0.699999988079071,
|
| 232 |
+
"rewards/chosen": 0.4612547755241394,
|
| 233 |
+
"rewards/margins": 0.084444060921669,
|
| 234 |
+
"rewards/rejected": 0.3768107295036316,
|
| 235 |
+
"step": 140
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.17,
|
| 239 |
+
"learning_rate": 4.932112504623876e-06,
|
| 240 |
+
"logits/chosen": -0.8811306953430176,
|
| 241 |
+
"logits/rejected": -1.0080692768096924,
|
| 242 |
+
"logps/chosen": -805.9840698242188,
|
| 243 |
+
"logps/rejected": -609.9173583984375,
|
| 244 |
+
"loss": 0.6289,
|
| 245 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 246 |
+
"rewards/chosen": 0.5022997260093689,
|
| 247 |
+
"rewards/margins": 0.16012665629386902,
|
| 248 |
+
"rewards/rejected": 0.34217312932014465,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.18,
|
| 253 |
+
"learning_rate": 4.907749318497991e-06,
|
| 254 |
+
"logits/chosen": -0.9709693193435669,
|
| 255 |
+
"logits/rejected": -1.1685694456100464,
|
| 256 |
+
"logps/chosen": -752.7962646484375,
|
| 257 |
+
"logps/rejected": -603.7000122070312,
|
| 258 |
+
"loss": 0.6427,
|
| 259 |
+
"rewards/accuracies": 0.675000011920929,
|
| 260 |
+
"rewards/chosen": 0.4435661733150482,
|
| 261 |
+
"rewards/margins": 0.13528670370578766,
|
| 262 |
+
"rewards/rejected": 0.30827948451042175,
|
| 263 |
+
"step": 160
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.19,
|
| 267 |
+
"learning_rate": 4.879737676511367e-06,
|
| 268 |
+
"logits/chosen": -0.9455941319465637,
|
| 269 |
+
"logits/rejected": -1.234893560409546,
|
| 270 |
+
"logps/chosen": -779.9085083007812,
|
| 271 |
+
"logps/rejected": -608.73291015625,
|
| 272 |
+
"loss": 0.6212,
|
| 273 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 274 |
+
"rewards/chosen": 0.5160066485404968,
|
| 275 |
+
"rewards/margins": 0.17969150841236115,
|
| 276 |
+
"rewards/rejected": 0.3363151252269745,
|
| 277 |
+
"step": 170
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.2,
|
| 281 |
+
"learning_rate": 4.848120024627372e-06,
|
| 282 |
+
"logits/chosen": -1.0280869007110596,
|
| 283 |
+
"logits/rejected": -1.0985267162322998,
|
| 284 |
+
"logps/chosen": -857.74609375,
|
| 285 |
+
"logps/rejected": -563.0875244140625,
|
| 286 |
+
"loss": 0.6145,
|
| 287 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 288 |
+
"rewards/chosen": 0.5581072568893433,
|
| 289 |
+
"rewards/margins": 0.19067201018333435,
|
| 290 |
+
"rewards/rejected": 0.36743518710136414,
|
| 291 |
+
"step": 180
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.21,
|
| 295 |
+
"learning_rate": 4.812944272986166e-06,
|
| 296 |
+
"logits/chosen": -1.0628365278244019,
|
| 297 |
+
"logits/rejected": -1.268007516860962,
|
| 298 |
+
"logps/chosen": -884.31884765625,
|
| 299 |
+
"logps/rejected": -716.6842651367188,
|
| 300 |
+
"loss": 0.6429,
|
| 301 |
+
"rewards/accuracies": 0.699999988079071,
|
| 302 |
+
"rewards/chosen": 0.5645692944526672,
|
| 303 |
+
"rewards/margins": 0.12494896352291107,
|
| 304 |
+
"rewards/rejected": 0.43962034583091736,
|
| 305 |
+
"step": 190
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.22,
|
| 309 |
+
"learning_rate": 4.774263723306599e-06,
|
| 310 |
+
"logits/chosen": -1.1383074522018433,
|
| 311 |
+
"logits/rejected": -1.194709062576294,
|
| 312 |
+
"logps/chosen": -809.2498168945312,
|
| 313 |
+
"logps/rejected": -548.3192138671875,
|
| 314 |
+
"loss": 0.6079,
|
| 315 |
+
"rewards/accuracies": 0.699999988079071,
|
| 316 |
+
"rewards/chosen": 0.5740378499031067,
|
| 317 |
+
"rewards/margins": 0.20076331496238708,
|
| 318 |
+
"rewards/rejected": 0.3732745051383972,
|
| 319 |
+
"step": 200
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.22,
|
| 323 |
+
"eval_logits/chosen": -1.0529733896255493,
|
| 324 |
+
"eval_logits/rejected": -1.2187565565109253,
|
| 325 |
+
"eval_logps/chosen": -794.3279418945312,
|
| 326 |
+
"eval_logps/rejected": -642.638916015625,
|
| 327 |
+
"eval_loss": 0.6462566256523132,
|
| 328 |
+
"eval_rewards/accuracies": 0.6507537961006165,
|
| 329 |
+
"eval_rewards/chosen": 0.5613374710083008,
|
| 330 |
+
"eval_rewards/margins": 0.12678806483745575,
|
| 331 |
+
"eval_rewards/rejected": 0.4345494508743286,
|
| 332 |
+
"eval_runtime": 182.1958,
|
| 333 |
+
"eval_samples_per_second": 2.184,
|
| 334 |
+
"eval_steps_per_second": 2.184,
|
| 335 |
+
"step": 200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 0.23,
|
| 339 |
+
"learning_rate": 4.732136988118259e-06,
|
| 340 |
+
"logits/chosen": -1.0591025352478027,
|
| 341 |
+
"logits/rejected": -1.1989389657974243,
|
| 342 |
+
"logps/chosen": -871.8209838867188,
|
| 343 |
+
"logps/rejected": -564.6995849609375,
|
| 344 |
+
"loss": 0.617,
|
| 345 |
+
"rewards/accuracies": 0.699999988079071,
|
| 346 |
+
"rewards/chosen": 0.5726070404052734,
|
| 347 |
+
"rewards/margins": 0.1896977722644806,
|
| 348 |
+
"rewards/rejected": 0.38290935754776,
|
| 349 |
+
"step": 210
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 0.25,
|
| 353 |
+
"learning_rate": 4.6866279019460744e-06,
|
| 354 |
+
"logits/chosen": -1.0191378593444824,
|
| 355 |
+
"logits/rejected": -1.0795801877975464,
|
| 356 |
+
"logps/chosen": -749.4663696289062,
|
| 357 |
+
"logps/rejected": -598.59619140625,
|
| 358 |
+
"loss": 0.6369,
|
| 359 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 360 |
+
"rewards/chosen": 0.5931397676467896,
|
| 361 |
+
"rewards/margins": 0.16538675129413605,
|
| 362 |
+
"rewards/rejected": 0.4277530312538147,
|
| 363 |
+
"step": 220
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.26,
|
| 367 |
+
"learning_rate": 4.637805424582033e-06,
|
| 368 |
+
"logits/chosen": -1.0304447412490845,
|
| 369 |
+
"logits/rejected": -1.1981227397918701,
|
| 370 |
+
"logps/chosen": -828.51171875,
|
| 371 |
+
"logps/rejected": -623.6228637695312,
|
| 372 |
+
"loss": 0.6275,
|
| 373 |
+
"rewards/accuracies": 0.75,
|
| 374 |
+
"rewards/chosen": 0.6149128079414368,
|
| 375 |
+
"rewards/margins": 0.18098048865795135,
|
| 376 |
+
"rewards/rejected": 0.4339323043823242,
|
| 377 |
+
"step": 230
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.27,
|
| 381 |
+
"learning_rate": 4.585743536590599e-06,
|
| 382 |
+
"logits/chosen": -1.041161298751831,
|
| 383 |
+
"logits/rejected": -1.301321268081665,
|
| 384 |
+
"logps/chosen": -936.7018432617188,
|
| 385 |
+
"logps/rejected": -561.6809692382812,
|
| 386 |
+
"loss": 0.5616,
|
| 387 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 388 |
+
"rewards/chosen": 0.6473914384841919,
|
| 389 |
+
"rewards/margins": 0.3042447566986084,
|
| 390 |
+
"rewards/rejected": 0.3431466221809387,
|
| 391 |
+
"step": 240
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.28,
|
| 395 |
+
"learning_rate": 4.530521127206173e-06,
|
| 396 |
+
"logits/chosen": -0.9962241053581238,
|
| 397 |
+
"logits/rejected": -1.1600590944290161,
|
| 398 |
+
"logps/chosen": -969.3317260742188,
|
| 399 |
+
"logps/rejected": -634.9935302734375,
|
| 400 |
+
"loss": 0.5807,
|
| 401 |
+
"rewards/accuracies": 0.800000011920929,
|
| 402 |
+
"rewards/chosen": 0.6946474313735962,
|
| 403 |
+
"rewards/margins": 0.2879478335380554,
|
| 404 |
+
"rewards/rejected": 0.4066995680332184,
|
| 405 |
+
"step": 250
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.29,
|
| 409 |
+
"learning_rate": 4.472221874792454e-06,
|
| 410 |
+
"logits/chosen": -1.0495914220809937,
|
| 411 |
+
"logits/rejected": -1.218639612197876,
|
| 412 |
+
"logps/chosen": -753.663818359375,
|
| 413 |
+
"logps/rejected": -488.46539306640625,
|
| 414 |
+
"loss": 0.5902,
|
| 415 |
+
"rewards/accuracies": 0.824999988079071,
|
| 416 |
+
"rewards/chosen": 0.5778111219406128,
|
| 417 |
+
"rewards/margins": 0.25250428915023804,
|
| 418 |
+
"rewards/rejected": 0.32530683279037476,
|
| 419 |
+
"step": 260
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 0.3,
|
| 423 |
+
"learning_rate": 4.410934120044838e-06,
|
| 424 |
+
"logits/chosen": -1.087903618812561,
|
| 425 |
+
"logits/rejected": -1.208064317703247,
|
| 426 |
+
"logps/chosen": -869.251953125,
|
| 427 |
+
"logps/rejected": -616.2910766601562,
|
| 428 |
+
"loss": 0.617,
|
| 429 |
+
"rewards/accuracies": 0.675000011920929,
|
| 430 |
+
"rewards/chosen": 0.6315535306930542,
|
| 431 |
+
"rewards/margins": 0.21421785652637482,
|
| 432 |
+
"rewards/rejected": 0.4173356592655182,
|
| 433 |
+
"step": 270
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.31,
|
| 437 |
+
"learning_rate": 4.346750732128023e-06,
|
| 438 |
+
"logits/chosen": -0.9508038759231567,
|
| 439 |
+
"logits/rejected": -1.128430724143982,
|
| 440 |
+
"logps/chosen": -715.2760009765625,
|
| 441 |
+
"logps/rejected": -632.925048828125,
|
| 442 |
+
"loss": 0.6425,
|
| 443 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 444 |
+
"rewards/chosen": 0.6043787002563477,
|
| 445 |
+
"rewards/margins": 0.15197435021400452,
|
| 446 |
+
"rewards/rejected": 0.45240435004234314,
|
| 447 |
+
"step": 280
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.32,
|
| 451 |
+
"learning_rate": 4.279768967951605e-06,
|
| 452 |
+
"logits/chosen": -1.0877809524536133,
|
| 453 |
+
"logits/rejected": -1.196732997894287,
|
| 454 |
+
"logps/chosen": -732.0111694335938,
|
| 455 |
+
"logps/rejected": -645.47998046875,
|
| 456 |
+
"loss": 0.5899,
|
| 457 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 458 |
+
"rewards/chosen": 0.5990191698074341,
|
| 459 |
+
"rewards/margins": 0.24647493660449982,
|
| 460 |
+
"rewards/rejected": 0.35254424810409546,
|
| 461 |
+
"step": 290
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.33,
|
| 465 |
+
"learning_rate": 4.210090324796965e-06,
|
| 466 |
+
"logits/chosen": -1.1224285364151,
|
| 467 |
+
"logits/rejected": -1.2125352621078491,
|
| 468 |
+
"logps/chosen": -701.78173828125,
|
| 469 |
+
"logps/rejected": -531.5074462890625,
|
| 470 |
+
"loss": 0.6029,
|
| 471 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 472 |
+
"rewards/chosen": 0.5454891324043274,
|
| 473 |
+
"rewards/margins": 0.2352382391691208,
|
| 474 |
+
"rewards/rejected": 0.3102509081363678,
|
| 475 |
+
"step": 300
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.33,
|
| 479 |
+
"eval_logits/chosen": -1.0366891622543335,
|
| 480 |
+
"eval_logits/rejected": -1.1964892148971558,
|
| 481 |
+
"eval_logps/chosen": -793.2078247070312,
|
| 482 |
+
"eval_logps/rejected": -644.8512573242188,
|
| 483 |
+
"eval_loss": 0.6359822750091553,
|
| 484 |
+
"eval_rewards/accuracies": 0.6557788848876953,
|
| 485 |
+
"eval_rewards/chosen": 0.5725387930870056,
|
| 486 |
+
"eval_rewards/margins": 0.16011320054531097,
|
| 487 |
+
"eval_rewards/rejected": 0.4124256670475006,
|
| 488 |
+
"eval_runtime": 182.0794,
|
| 489 |
+
"eval_samples_per_second": 2.186,
|
| 490 |
+
"eval_steps_per_second": 2.186,
|
| 491 |
+
"step": 300
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.35,
|
| 495 |
+
"learning_rate": 4.137820386518716e-06,
|
| 496 |
+
"logits/chosen": -1.007893681526184,
|
| 497 |
+
"logits/rejected": -1.2096047401428223,
|
| 498 |
+
"logps/chosen": -871.7268676757812,
|
| 499 |
+
"logps/rejected": -572.1101684570312,
|
| 500 |
+
"loss": 0.5544,
|
| 501 |
+
"rewards/accuracies": 0.824999988079071,
|
| 502 |
+
"rewards/chosen": 0.6873822808265686,
|
| 503 |
+
"rewards/margins": 0.35040804743766785,
|
| 504 |
+
"rewards/rejected": 0.33697420358657837,
|
| 505 |
+
"step": 310
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.36,
|
| 509 |
+
"learning_rate": 4.063068663553778e-06,
|
| 510 |
+
"logits/chosen": -1.0717604160308838,
|
| 511 |
+
"logits/rejected": -1.252614974975586,
|
| 512 |
+
"logps/chosen": -747.0122680664062,
|
| 513 |
+
"logps/rejected": -644.8168334960938,
|
| 514 |
+
"loss": 0.5943,
|
| 515 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 516 |
+
"rewards/chosen": 0.6160334944725037,
|
| 517 |
+
"rewards/margins": 0.25227323174476624,
|
| 518 |
+
"rewards/rejected": 0.36376023292541504,
|
| 519 |
+
"step": 320
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.37,
|
| 523 |
+
"learning_rate": 3.9859484269805215e-06,
|
| 524 |
+
"logits/chosen": -1.1249014139175415,
|
| 525 |
+
"logits/rejected": -1.2282134294509888,
|
| 526 |
+
"logps/chosen": -852.0465087890625,
|
| 527 |
+
"logps/rejected": -594.2730712890625,
|
| 528 |
+
"loss": 0.6014,
|
| 529 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 530 |
+
"rewards/chosen": 0.6664583683013916,
|
| 531 |
+
"rewards/margins": 0.2528621554374695,
|
| 532 |
+
"rewards/rejected": 0.4135962128639221,
|
| 533 |
+
"step": 330
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.38,
|
| 537 |
+
"learning_rate": 3.906576536879416e-06,
|
| 538 |
+
"logits/chosen": -1.0270237922668457,
|
| 539 |
+
"logits/rejected": -1.242011308670044,
|
| 540 |
+
"logps/chosen": -827.2908325195312,
|
| 541 |
+
"logps/rejected": -647.3381958007812,
|
| 542 |
+
"loss": 0.5662,
|
| 543 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 544 |
+
"rewards/chosen": 0.6910017728805542,
|
| 545 |
+
"rewards/margins": 0.3151417672634125,
|
| 546 |
+
"rewards/rejected": 0.3758600056171417,
|
| 547 |
+
"step": 340
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 0.39,
|
| 551 |
+
"learning_rate": 3.825073265255271e-06,
|
| 552 |
+
"logits/chosen": -1.1385161876678467,
|
| 553 |
+
"logits/rejected": -1.1466325521469116,
|
| 554 |
+
"logps/chosen": -724.5143432617188,
|
| 555 |
+
"logps/rejected": -692.8756103515625,
|
| 556 |
+
"loss": 0.6208,
|
| 557 |
+
"rewards/accuracies": 0.699999988079071,
|
| 558 |
+
"rewards/chosen": 0.6377968192100525,
|
| 559 |
+
"rewards/margins": 0.20125746726989746,
|
| 560 |
+
"rewards/rejected": 0.43653935194015503,
|
| 561 |
+
"step": 350
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.4,
|
| 565 |
+
"learning_rate": 3.7415621137894055e-06,
|
| 566 |
+
"logits/chosen": -1.1019694805145264,
|
| 567 |
+
"logits/rejected": -1.3558331727981567,
|
| 568 |
+
"logps/chosen": -752.0270385742188,
|
| 569 |
+
"logps/rejected": -574.9373779296875,
|
| 570 |
+
"loss": 0.5653,
|
| 571 |
+
"rewards/accuracies": 0.675000011920929,
|
| 572 |
+
"rewards/chosen": 0.6500095725059509,
|
| 573 |
+
"rewards/margins": 0.3266737759113312,
|
| 574 |
+
"rewards/rejected": 0.3233358561992645,
|
| 575 |
+
"step": 360
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.41,
|
| 579 |
+
"learning_rate": 3.656169626697889e-06,
|
| 580 |
+
"logits/chosen": -1.1742336750030518,
|
| 581 |
+
"logits/rejected": -1.2145332098007202,
|
| 582 |
+
"logps/chosen": -733.7666015625,
|
| 583 |
+
"logps/rejected": -577.3091430664062,
|
| 584 |
+
"loss": 0.6117,
|
| 585 |
+
"rewards/accuracies": 0.699999988079071,
|
| 586 |
+
"rewards/chosen": 0.6057840585708618,
|
| 587 |
+
"rewards/margins": 0.2176964282989502,
|
| 588 |
+
"rewards/rejected": 0.3880876302719116,
|
| 589 |
+
"step": 370
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 0.42,
|
| 593 |
+
"learning_rate": 3.5690251989794443e-06,
|
| 594 |
+
"logits/chosen": -1.1451467275619507,
|
| 595 |
+
"logits/rejected": -1.4090936183929443,
|
| 596 |
+
"logps/chosen": -791.69287109375,
|
| 597 |
+
"logps/rejected": -664.351318359375,
|
| 598 |
+
"loss": 0.5472,
|
| 599 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 600 |
+
"rewards/chosen": 0.7302919626235962,
|
| 601 |
+
"rewards/margins": 0.39029377698898315,
|
| 602 |
+
"rewards/rejected": 0.33999815583229065,
|
| 603 |
+
"step": 380
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.43,
|
| 607 |
+
"learning_rate": 3.480260880343565e-06,
|
| 608 |
+
"logits/chosen": -1.1370658874511719,
|
| 609 |
+
"logits/rejected": -1.268176555633545,
|
| 610 |
+
"logps/chosen": -704.5770263671875,
|
| 611 |
+
"logps/rejected": -549.7819213867188,
|
| 612 |
+
"loss": 0.5728,
|
| 613 |
+
"rewards/accuracies": 0.699999988079071,
|
| 614 |
+
"rewards/chosen": 0.665199875831604,
|
| 615 |
+
"rewards/margins": 0.32814812660217285,
|
| 616 |
+
"rewards/rejected": 0.3370516896247864,
|
| 617 |
+
"step": 390
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.45,
|
| 621 |
+
"learning_rate": 3.390011175115956e-06,
|
| 622 |
+
"logits/chosen": -1.0938374996185303,
|
| 623 |
+
"logits/rejected": -1.208974003791809,
|
| 624 |
+
"logps/chosen": -662.2012939453125,
|
| 625 |
+
"logps/rejected": -570.6355590820312,
|
| 626 |
+
"loss": 0.6123,
|
| 627 |
+
"rewards/accuracies": 0.699999988079071,
|
| 628 |
+
"rewards/chosen": 0.6198933124542236,
|
| 629 |
+
"rewards/margins": 0.2259746491909027,
|
| 630 |
+
"rewards/rejected": 0.39391860365867615,
|
| 631 |
+
"step": 400
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 0.45,
|
| 635 |
+
"eval_logits/chosen": -1.1220430135726929,
|
| 636 |
+
"eval_logits/rejected": -1.265822172164917,
|
| 637 |
+
"eval_logps/chosen": -787.7749633789062,
|
| 638 |
+
"eval_logps/rejected": -641.9633178710938,
|
| 639 |
+
"eval_loss": 0.6290538907051086,
|
| 640 |
+
"eval_rewards/accuracies": 0.660804033279419,
|
| 641 |
+
"eval_rewards/chosen": 0.6268669962882996,
|
| 642 |
+
"eval_rewards/margins": 0.18556177616119385,
|
| 643 |
+
"eval_rewards/rejected": 0.4413052201271057,
|
| 644 |
+
"eval_runtime": 182.0502,
|
| 645 |
+
"eval_samples_per_second": 2.186,
|
| 646 |
+
"eval_steps_per_second": 2.186,
|
| 647 |
+
"step": 400
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.46,
|
| 651 |
+
"learning_rate": 3.298412838424503e-06,
|
| 652 |
+
"logits/chosen": -1.1896421909332275,
|
| 653 |
+
"logits/rejected": -1.2548556327819824,
|
| 654 |
+
"logps/chosen": -818.2477416992188,
|
| 655 |
+
"logps/rejected": -607.974365234375,
|
| 656 |
+
"loss": 0.5652,
|
| 657 |
+
"rewards/accuracies": 0.75,
|
| 658 |
+
"rewards/chosen": 0.6872069239616394,
|
| 659 |
+
"rewards/margins": 0.32436543703079224,
|
| 660 |
+
"rewards/rejected": 0.36284154653549194,
|
| 661 |
+
"step": 410
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.47,
|
| 665 |
+
"learning_rate": 3.205604668974607e-06,
|
| 666 |
+
"logits/chosen": -1.0528907775878906,
|
| 667 |
+
"logits/rejected": -1.1407417058944702,
|
| 668 |
+
"logps/chosen": -798.6719970703125,
|
| 669 |
+
"logps/rejected": -551.5494995117188,
|
| 670 |
+
"loss": 0.5562,
|
| 671 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 672 |
+
"rewards/chosen": 0.7110682725906372,
|
| 673 |
+
"rewards/margins": 0.3393869996070862,
|
| 674 |
+
"rewards/rejected": 0.37168124318122864,
|
| 675 |
+
"step": 420
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.48,
|
| 679 |
+
"learning_rate": 3.111727298727888e-06,
|
| 680 |
+
"logits/chosen": -1.0838218927383423,
|
| 681 |
+
"logits/rejected": -1.3229072093963623,
|
| 682 |
+
"logps/chosen": -683.1719970703125,
|
| 683 |
+
"logps/rejected": -585.8663330078125,
|
| 684 |
+
"loss": 0.6278,
|
| 685 |
+
"rewards/accuracies": 0.699999988079071,
|
| 686 |
+
"rewards/chosen": 0.5717045068740845,
|
| 687 |
+
"rewards/margins": 0.19018582999706268,
|
| 688 |
+
"rewards/rejected": 0.3815186619758606,
|
| 689 |
+
"step": 430
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.49,
|
| 693 |
+
"learning_rate": 3.0169229798029698e-06,
|
| 694 |
+
"logits/chosen": -1.1374232769012451,
|
| 695 |
+
"logits/rejected": -1.2010905742645264,
|
| 696 |
+
"logps/chosen": -860.7879638671875,
|
| 697 |
+
"logps/rejected": -513.562744140625,
|
| 698 |
+
"loss": 0.5396,
|
| 699 |
+
"rewards/accuracies": 0.75,
|
| 700 |
+
"rewards/chosen": 0.7182751893997192,
|
| 701 |
+
"rewards/margins": 0.4120573103427887,
|
| 702 |
+
"rewards/rejected": 0.30621784925460815,
|
| 703 |
+
"step": 440
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.5,
|
| 707 |
+
"learning_rate": 2.9213353689212337e-06,
|
| 708 |
+
"logits/chosen": -1.1559637784957886,
|
| 709 |
+
"logits/rejected": -1.3046997785568237,
|
| 710 |
+
"logps/chosen": -728.3955078125,
|
| 711 |
+
"logps/rejected": -552.3156127929688,
|
| 712 |
+
"loss": 0.5476,
|
| 713 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 714 |
+
"rewards/chosen": 0.6234858632087708,
|
| 715 |
+
"rewards/margins": 0.37362077832221985,
|
| 716 |
+
"rewards/rejected": 0.2498650997877121,
|
| 717 |
+
"step": 450
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.51,
|
| 721 |
+
"learning_rate": 2.8251093097241895e-06,
|
| 722 |
+
"logits/chosen": -1.0692123174667358,
|
| 723 |
+
"logits/rejected": -1.2266539335250854,
|
| 724 |
+
"logps/chosen": -792.2379150390625,
|
| 725 |
+
"logps/rejected": -661.500732421875,
|
| 726 |
+
"loss": 0.615,
|
| 727 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 728 |
+
"rewards/chosen": 0.6842690706253052,
|
| 729 |
+
"rewards/margins": 0.26619952917099,
|
| 730 |
+
"rewards/rejected": 0.4180695414543152,
|
| 731 |
+
"step": 460
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.52,
|
| 735 |
+
"learning_rate": 2.7283906132923104e-06,
|
| 736 |
+
"logits/chosen": -1.1081640720367432,
|
| 737 |
+
"logits/rejected": -1.196122407913208,
|
| 738 |
+
"logps/chosen": -861.2401123046875,
|
| 739 |
+
"logps/rejected": -630.8099365234375,
|
| 740 |
+
"loss": 0.5369,
|
| 741 |
+
"rewards/accuracies": 0.824999988079071,
|
| 742 |
+
"rewards/chosen": 0.7281755208969116,
|
| 743 |
+
"rewards/margins": 0.4001568853855133,
|
| 744 |
+
"rewards/rejected": 0.3280186355113983,
|
| 745 |
+
"step": 470
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.54,
|
| 749 |
+
"learning_rate": 2.6313258371978996e-06,
|
| 750 |
+
"logits/chosen": -1.0956523418426514,
|
| 751 |
+
"logits/rejected": -1.2124046087265015,
|
| 752 |
+
"logps/chosen": -806.178466796875,
|
| 753 |
+
"logps/rejected": -556.8995361328125,
|
| 754 |
+
"loss": 0.5704,
|
| 755 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 756 |
+
"rewards/chosen": 0.6232426762580872,
|
| 757 |
+
"rewards/margins": 0.3145415782928467,
|
| 758 |
+
"rewards/rejected": 0.3087010979652405,
|
| 759 |
+
"step": 480
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.55,
|
| 763 |
+
"learning_rate": 2.5340620634268167e-06,
|
| 764 |
+
"logits/chosen": -1.1129533052444458,
|
| 765 |
+
"logits/rejected": -1.3338663578033447,
|
| 766 |
+
"logps/chosen": -722.3619995117188,
|
| 767 |
+
"logps/rejected": -528.3717041015625,
|
| 768 |
+
"loss": 0.5459,
|
| 769 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 770 |
+
"rewards/chosen": 0.6257763504981995,
|
| 771 |
+
"rewards/margins": 0.37175193428993225,
|
| 772 |
+
"rewards/rejected": 0.2540244162082672,
|
| 773 |
+
"step": 490
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.56,
|
| 777 |
+
"learning_rate": 2.436746675505545e-06,
|
| 778 |
+
"logits/chosen": -1.0671157836914062,
|
| 779 |
+
"logits/rejected": -1.2726842164993286,
|
| 780 |
+
"logps/chosen": -709.8563842773438,
|
| 781 |
+
"logps/rejected": -538.7957763671875,
|
| 782 |
+
"loss": 0.5596,
|
| 783 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 784 |
+
"rewards/chosen": 0.6499138474464417,
|
| 785 |
+
"rewards/margins": 0.3205423951148987,
|
| 786 |
+
"rewards/rejected": 0.3293713927268982,
|
| 787 |
+
"step": 500
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.56,
|
| 791 |
+
"eval_logits/chosen": -1.085227131843567,
|
| 792 |
+
"eval_logits/rejected": -1.2330302000045776,
|
| 793 |
+
"eval_logps/chosen": -790.7927856445312,
|
| 794 |
+
"eval_logps/rejected": -646.7930297851562,
|
| 795 |
+
"eval_loss": 0.6230102181434631,
|
| 796 |
+
"eval_rewards/accuracies": 0.6683416962623596,
|
| 797 |
+
"eval_rewards/chosen": 0.5966897010803223,
|
| 798 |
+
"eval_rewards/margins": 0.20368172228336334,
|
| 799 |
+
"eval_rewards/rejected": 0.39300796389579773,
|
| 800 |
+
"eval_runtime": 182.069,
|
| 801 |
+
"eval_samples_per_second": 2.186,
|
| 802 |
+
"eval_steps_per_second": 2.186,
|
| 803 |
+
"step": 500
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 0.57,
|
| 807 |
+
"learning_rate": 2.3395271351713515e-06,
|
| 808 |
+
"logits/chosen": -1.0355545282363892,
|
| 809 |
+
"logits/rejected": -1.2773510217666626,
|
| 810 |
+
"logps/chosen": -846.6865234375,
|
| 811 |
+
"logps/rejected": -515.6051635742188,
|
| 812 |
+
"loss": 0.5052,
|
| 813 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 814 |
+
"rewards/chosen": 0.7559247016906738,
|
| 815 |
+
"rewards/margins": 0.48392027616500854,
|
| 816 |
+
"rewards/rejected": 0.2720043659210205,
|
| 817 |
+
"step": 510
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 0.58,
|
| 821 |
+
"learning_rate": 2.2425507589239154e-06,
|
| 822 |
+
"logits/chosen": -1.0901484489440918,
|
| 823 |
+
"logits/rejected": -1.2016217708587646,
|
| 824 |
+
"logps/chosen": -946.5360107421875,
|
| 825 |
+
"logps/rejected": -568.994384765625,
|
| 826 |
+
"loss": 0.519,
|
| 827 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 828 |
+
"rewards/chosen": 0.8060197830200195,
|
| 829 |
+
"rewards/margins": 0.48728522658348083,
|
| 830 |
+
"rewards/rejected": 0.3187345862388611,
|
| 831 |
+
"step": 520
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.59,
|
| 835 |
+
"learning_rate": 2.145964494797051e-06,
|
| 836 |
+
"logits/chosen": -1.0998233556747437,
|
| 837 |
+
"logits/rejected": -1.2087761163711548,
|
| 838 |
+
"logps/chosen": -728.5558471679688,
|
| 839 |
+
"logps/rejected": -671.7276611328125,
|
| 840 |
+
"loss": 0.6116,
|
| 841 |
+
"rewards/accuracies": 0.699999988079071,
|
| 842 |
+
"rewards/chosen": 0.6140581965446472,
|
| 843 |
+
"rewards/margins": 0.23174139857292175,
|
| 844 |
+
"rewards/rejected": 0.3823167681694031,
|
| 845 |
+
"step": 530
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 0.6,
|
| 849 |
+
"learning_rate": 2.049914699688762e-06,
|
| 850 |
+
"logits/chosen": -1.1990127563476562,
|
| 851 |
+
"logits/rejected": -1.2407617568969727,
|
| 852 |
+
"logps/chosen": -739.6600341796875,
|
| 853 |
+
"logps/rejected": -640.65380859375,
|
| 854 |
+
"loss": 0.5987,
|
| 855 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 856 |
+
"rewards/chosen": 0.6110137104988098,
|
| 857 |
+
"rewards/margins": 0.2484373301267624,
|
| 858 |
+
"rewards/rejected": 0.36257636547088623,
|
| 859 |
+
"step": 540
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 0.61,
|
| 863 |
+
"learning_rate": 1.954546917587033e-06,
|
| 864 |
+
"logits/chosen": -1.1063880920410156,
|
| 865 |
+
"logits/rejected": -1.2011101245880127,
|
| 866 |
+
"logps/chosen": -756.6212768554688,
|
| 867 |
+
"logps/rejected": -455.97979736328125,
|
| 868 |
+
"loss": 0.5123,
|
| 869 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 870 |
+
"rewards/chosen": 0.7002598643302917,
|
| 871 |
+
"rewards/margins": 0.47388529777526855,
|
| 872 |
+
"rewards/rejected": 0.22637462615966797,
|
| 873 |
+
"step": 550
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.62,
|
| 877 |
+
"learning_rate": 1.8600056590274355e-06,
|
| 878 |
+
"logits/chosen": -1.095252513885498,
|
| 879 |
+
"logits/rejected": -1.1555767059326172,
|
| 880 |
+
"logps/chosen": -730.5572509765625,
|
| 881 |
+
"logps/rejected": -615.13134765625,
|
| 882 |
+
"loss": 0.619,
|
| 883 |
+
"rewards/accuracies": 0.675000011920929,
|
| 884 |
+
"rewards/chosen": 0.6304242610931396,
|
| 885 |
+
"rewards/margins": 0.23751536011695862,
|
| 886 |
+
"rewards/rejected": 0.3929089605808258,
|
| 887 |
+
"step": 560
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 0.64,
|
| 891 |
+
"learning_rate": 1.766434182116708e-06,
|
| 892 |
+
"logits/chosen": -1.173472285270691,
|
| 893 |
+
"logits/rejected": -1.1808496713638306,
|
| 894 |
+
"logps/chosen": -754.0181884765625,
|
| 895 |
+
"logps/rejected": -660.9012451171875,
|
| 896 |
+
"loss": 0.6232,
|
| 897 |
+
"rewards/accuracies": 0.675000011920929,
|
| 898 |
+
"rewards/chosen": 0.5536493062973022,
|
| 899 |
+
"rewards/margins": 0.21175837516784668,
|
| 900 |
+
"rewards/rejected": 0.34189099073410034,
|
| 901 |
+
"step": 570
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 0.65,
|
| 905 |
+
"learning_rate": 1.6739742754541515e-06,
|
| 906 |
+
"logits/chosen": -0.9952909350395203,
|
| 907 |
+
"logits/rejected": -1.2373504638671875,
|
| 908 |
+
"logps/chosen": -740.7738037109375,
|
| 909 |
+
"logps/rejected": -658.3294067382812,
|
| 910 |
+
"loss": 0.57,
|
| 911 |
+
"rewards/accuracies": 0.75,
|
| 912 |
+
"rewards/chosen": 0.6436474323272705,
|
| 913 |
+
"rewards/margins": 0.3406828045845032,
|
| 914 |
+
"rewards/rejected": 0.30296462774276733,
|
| 915 |
+
"step": 580
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 0.66,
|
| 919 |
+
"learning_rate": 1.582766043279752e-06,
|
| 920 |
+
"logits/chosen": -1.0711066722869873,
|
| 921 |
+
"logits/rejected": -1.2044506072998047,
|
| 922 |
+
"logps/chosen": -700.0191650390625,
|
| 923 |
+
"logps/rejected": -573.0767211914062,
|
| 924 |
+
"loss": 0.5885,
|
| 925 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 926 |
+
"rewards/chosen": 0.5518115758895874,
|
| 927 |
+
"rewards/margins": 0.28940147161483765,
|
| 928 |
+
"rewards/rejected": 0.26241010427474976,
|
| 929 |
+
"step": 590
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.67,
|
| 933 |
+
"learning_rate": 1.4929476931746167e-06,
|
| 934 |
+
"logits/chosen": -0.9843810796737671,
|
| 935 |
+
"logits/rejected": -1.1629724502563477,
|
| 936 |
+
"logps/chosen": -815.2390747070312,
|
| 937 |
+
"logps/rejected": -556.5223999023438,
|
| 938 |
+
"loss": 0.5382,
|
| 939 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 940 |
+
"rewards/chosen": 0.6965817213058472,
|
| 941 |
+
"rewards/margins": 0.40665000677108765,
|
| 942 |
+
"rewards/rejected": 0.2899317145347595,
|
| 943 |
+
"step": 600
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 0.67,
|
| 947 |
+
"eval_logits/chosen": -1.054700493812561,
|
| 948 |
+
"eval_logits/rejected": -1.2033686637878418,
|
| 949 |
+
"eval_logps/chosen": -793.2485961914062,
|
| 950 |
+
"eval_logps/rejected": -650.0925903320312,
|
| 951 |
+
"eval_loss": 0.6198973059654236,
|
| 952 |
+
"eval_rewards/accuracies": 0.6708542704582214,
|
| 953 |
+
"eval_rewards/chosen": 0.5721314549446106,
|
| 954 |
+
"eval_rewards/margins": 0.21211864054203033,
|
| 955 |
+
"eval_rewards/rejected": 0.36001279950141907,
|
| 956 |
+
"eval_runtime": 181.9771,
|
| 957 |
+
"eval_samples_per_second": 2.187,
|
| 958 |
+
"eval_steps_per_second": 2.187,
|
| 959 |
+
"step": 600
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.68,
|
| 963 |
+
"learning_rate": 1.4046553266354126e-06,
|
| 964 |
+
"logits/chosen": -1.1579875946044922,
|
| 965 |
+
"logits/rejected": -1.1627570390701294,
|
| 966 |
+
"logps/chosen": -710.740966796875,
|
| 967 |
+
"logps/rejected": -705.4386596679688,
|
| 968 |
+
"loss": 0.651,
|
| 969 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 970 |
+
"rewards/chosen": 0.5279535055160522,
|
| 971 |
+
"rewards/margins": 0.15107524394989014,
|
| 972 |
+
"rewards/rejected": 0.3768783211708069,
|
| 973 |
+
"step": 610
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 0.69,
|
| 977 |
+
"learning_rate": 1.318022732840141e-06,
|
| 978 |
+
"logits/chosen": -1.082596778869629,
|
| 979 |
+
"logits/rejected": -1.2048799991607666,
|
| 980 |
+
"logps/chosen": -773.497802734375,
|
| 981 |
+
"logps/rejected": -556.2584838867188,
|
| 982 |
+
"loss": 0.5922,
|
| 983 |
+
"rewards/accuracies": 0.800000011920929,
|
| 984 |
+
"rewards/chosen": 0.5189529657363892,
|
| 985 |
+
"rewards/margins": 0.2554023265838623,
|
| 986 |
+
"rewards/rejected": 0.26355063915252686,
|
| 987 |
+
"step": 620
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 0.7,
|
| 991 |
+
"learning_rate": 1.2331811859177722e-06,
|
| 992 |
+
"logits/chosen": -1.1044152975082397,
|
| 993 |
+
"logits/rejected": -1.3350199460983276,
|
| 994 |
+
"logps/chosen": -763.0442504882812,
|
| 995 |
+
"logps/rejected": -605.748046875,
|
| 996 |
+
"loss": 0.5383,
|
| 997 |
+
"rewards/accuracies": 0.75,
|
| 998 |
+
"rewards/chosen": 0.6514481902122498,
|
| 999 |
+
"rewards/margins": 0.42245060205459595,
|
| 1000 |
+
"rewards/rejected": 0.2289975881576538,
|
| 1001 |
+
"step": 630
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 0.71,
|
| 1005 |
+
"learning_rate": 1.150259246028921e-06,
|
| 1006 |
+
"logits/chosen": -0.9752397537231445,
|
| 1007 |
+
"logits/rejected": -1.1639258861541748,
|
| 1008 |
+
"logps/chosen": -736.8223266601562,
|
| 1009 |
+
"logps/rejected": -610.3005981445312,
|
| 1010 |
+
"loss": 0.5711,
|
| 1011 |
+
"rewards/accuracies": 0.75,
|
| 1012 |
+
"rewards/chosen": 0.6042425036430359,
|
| 1013 |
+
"rewards/margins": 0.3154502809047699,
|
| 1014 |
+
"rewards/rejected": 0.288792222738266,
|
| 1015 |
+
"step": 640
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 0.72,
|
| 1019 |
+
"learning_rate": 1.0693825645589887e-06,
|
| 1020 |
+
"logits/chosen": -0.9845565557479858,
|
| 1021 |
+
"logits/rejected": -1.205209732055664,
|
| 1022 |
+
"logps/chosen": -778.3556518554688,
|
| 1023 |
+
"logps/rejected": -546.72265625,
|
| 1024 |
+
"loss": 0.5316,
|
| 1025 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1026 |
+
"rewards/chosen": 0.6856497526168823,
|
| 1027 |
+
"rewards/margins": 0.4366929531097412,
|
| 1028 |
+
"rewards/rejected": 0.24895676970481873,
|
| 1029 |
+
"step": 650
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.74,
|
| 1033 |
+
"learning_rate": 9.9067369371897e-07,
|
| 1034 |
+
"logits/chosen": -1.1334224939346313,
|
| 1035 |
+
"logits/rejected": -1.261704683303833,
|
| 1036 |
+
"logps/chosen": -795.8439331054688,
|
| 1037 |
+
"logps/rejected": -630.4856567382812,
|
| 1038 |
+
"loss": 0.5529,
|
| 1039 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1040 |
+
"rewards/chosen": 0.6260807514190674,
|
| 1041 |
+
"rewards/margins": 0.3732090890407562,
|
| 1042 |
+
"rewards/rejected": 0.25287163257598877,
|
| 1043 |
+
"step": 660
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 0.75,
|
| 1047 |
+
"learning_rate": 9.14251900842432e-07,
|
| 1048 |
+
"logits/chosen": -1.0554249286651611,
|
| 1049 |
+
"logits/rejected": -1.1402459144592285,
|
| 1050 |
+
"logps/chosen": -697.9842529296875,
|
| 1051 |
+
"logps/rejected": -549.4150390625,
|
| 1052 |
+
"loss": 0.5811,
|
| 1053 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1054 |
+
"rewards/chosen": 0.530659556388855,
|
| 1055 |
+
"rewards/margins": 0.29245835542678833,
|
| 1056 |
+
"rewards/rejected": 0.23820118606090546,
|
| 1057 |
+
"step": 670
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 0.76,
|
| 1061 |
+
"learning_rate": 8.402329876600462e-07,
|
| 1062 |
+
"logits/chosen": -1.0341722965240479,
|
| 1063 |
+
"logits/rejected": -1.126814603805542,
|
| 1064 |
+
"logps/chosen": -864.65966796875,
|
| 1065 |
+
"logps/rejected": -632.652099609375,
|
| 1066 |
+
"loss": 0.5377,
|
| 1067 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1068 |
+
"rewards/chosen": 0.72022545337677,
|
| 1069 |
+
"rewards/margins": 0.4036920964717865,
|
| 1070 |
+
"rewards/rejected": 0.3165333867073059,
|
| 1071 |
+
"step": 680
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 0.77,
|
| 1075 |
+
"learning_rate": 7.687291148255527e-07,
|
| 1076 |
+
"logits/chosen": -1.0247344970703125,
|
| 1077 |
+
"logits/rejected": -1.1219924688339233,
|
| 1078 |
+
"logps/chosen": -944.3568115234375,
|
| 1079 |
+
"logps/rejected": -662.272216796875,
|
| 1080 |
+
"loss": 0.5183,
|
| 1081 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1082 |
+
"rewards/chosen": 0.8120288848876953,
|
| 1083 |
+
"rewards/margins": 0.4406498968601227,
|
| 1084 |
+
"rewards/rejected": 0.37137895822525024,
|
| 1085 |
+
"step": 690
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 0.78,
|
| 1089 |
+
"learning_rate": 6.998486319590323e-07,
|
| 1090 |
+
"logits/chosen": -1.0355504751205444,
|
| 1091 |
+
"logits/rejected": -1.0869207382202148,
|
| 1092 |
+
"logps/chosen": -768.5548095703125,
|
| 1093 |
+
"logps/rejected": -638.7791748046875,
|
| 1094 |
+
"loss": 0.5952,
|
| 1095 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1096 |
+
"rewards/chosen": 0.6144096851348877,
|
| 1097 |
+
"rewards/margins": 0.27622345089912415,
|
| 1098 |
+
"rewards/rejected": 0.33818623423576355,
|
| 1099 |
+
"step": 700
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 0.78,
|
| 1103 |
+
"eval_logits/chosen": -1.0323765277862549,
|
| 1104 |
+
"eval_logits/rejected": -1.182665228843689,
|
| 1105 |
+
"eval_logps/chosen": -794.96044921875,
|
| 1106 |
+
"eval_logps/rejected": -652.0419921875,
|
| 1107 |
+
"eval_loss": 0.6185528635978699,
|
| 1108 |
+
"eval_rewards/accuracies": 0.6783919334411621,
|
| 1109 |
+
"eval_rewards/chosen": 0.5550140738487244,
|
| 1110 |
+
"eval_rewards/margins": 0.21449486911296844,
|
| 1111 |
+
"eval_rewards/rejected": 0.3405191898345947,
|
| 1112 |
+
"eval_runtime": 182.1196,
|
| 1113 |
+
"eval_samples_per_second": 2.185,
|
| 1114 |
+
"eval_steps_per_second": 2.185,
|
| 1115 |
+
"step": 700
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.79,
|
| 1119 |
+
"learning_rate": 6.336959134650278e-07,
|
| 1120 |
+
"logits/chosen": -1.1061818599700928,
|
| 1121 |
+
"logits/rejected": -1.1679563522338867,
|
| 1122 |
+
"logps/chosen": -823.1553955078125,
|
| 1123 |
+
"logps/rejected": -556.3447875976562,
|
| 1124 |
+
"loss": 0.5435,
|
| 1125 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1126 |
+
"rewards/chosen": 0.6446937322616577,
|
| 1127 |
+
"rewards/margins": 0.3869627118110657,
|
| 1128 |
+
"rewards/rejected": 0.2577310800552368,
|
| 1129 |
+
"step": 710
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.8,
|
| 1133 |
+
"learning_rate": 5.703712003742965e-07,
|
| 1134 |
+
"logits/chosen": -1.013837456703186,
|
| 1135 |
+
"logits/rejected": -1.2135231494903564,
|
| 1136 |
+
"logps/chosen": -790.27490234375,
|
| 1137 |
+
"logps/rejected": -591.6689453125,
|
| 1138 |
+
"loss": 0.5499,
|
| 1139 |
+
"rewards/accuracies": 0.75,
|
| 1140 |
+
"rewards/chosen": 0.6053417325019836,
|
| 1141 |
+
"rewards/margins": 0.37499576807022095,
|
| 1142 |
+
"rewards/rejected": 0.2303459644317627,
|
| 1143 |
+
"step": 720
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.81,
|
| 1147 |
+
"learning_rate": 5.099704484488569e-07,
|
| 1148 |
+
"logits/chosen": -1.0917243957519531,
|
| 1149 |
+
"logits/rejected": -1.1174938678741455,
|
| 1150 |
+
"logps/chosen": -800.6257934570312,
|
| 1151 |
+
"logps/rejected": -576.2686767578125,
|
| 1152 |
+
"loss": 0.5762,
|
| 1153 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1154 |
+
"rewards/chosen": 0.5678650736808777,
|
| 1155 |
+
"rewards/margins": 0.30998459458351135,
|
| 1156 |
+
"rewards/rejected": 0.25788044929504395,
|
| 1157 |
+
"step": 730
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.82,
|
| 1161 |
+
"learning_rate": 4.525851827804881e-07,
|
| 1162 |
+
"logits/chosen": -1.0324584245681763,
|
| 1163 |
+
"logits/rejected": -1.2507680654525757,
|
| 1164 |
+
"logps/chosen": -849.1965942382812,
|
| 1165 |
+
"logps/rejected": -620.4736328125,
|
| 1166 |
+
"loss": 0.5563,
|
| 1167 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1168 |
+
"rewards/chosen": 0.6641701459884644,
|
| 1169 |
+
"rewards/margins": 0.3699476420879364,
|
| 1170 |
+
"rewards/rejected": 0.2942224442958832,
|
| 1171 |
+
"step": 740
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.84,
|
| 1175 |
+
"learning_rate": 3.983023591030113e-07,
|
| 1176 |
+
"logits/chosen": -1.0120137929916382,
|
| 1177 |
+
"logits/rejected": -1.2262170314788818,
|
| 1178 |
+
"logps/chosen": -834.7081909179688,
|
| 1179 |
+
"logps/rejected": -564.6629638671875,
|
| 1180 |
+
"loss": 0.5074,
|
| 1181 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1182 |
+
"rewards/chosen": 0.6688768267631531,
|
| 1183 |
+
"rewards/margins": 0.48412173986434937,
|
| 1184 |
+
"rewards/rejected": 0.1847551316022873,
|
| 1185 |
+
"step": 750
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.85,
|
| 1189 |
+
"learning_rate": 3.472042320285071e-07,
|
| 1190 |
+
"logits/chosen": -1.0783493518829346,
|
| 1191 |
+
"logits/rejected": -1.2198795080184937,
|
| 1192 |
+
"logps/chosen": -753.7183837890625,
|
| 1193 |
+
"logps/rejected": -542.9868774414062,
|
| 1194 |
+
"loss": 0.5736,
|
| 1195 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1196 |
+
"rewards/chosen": 0.5568192005157471,
|
| 1197 |
+
"rewards/margins": 0.3265761733055115,
|
| 1198 |
+
"rewards/rejected": 0.2302430123090744,
|
| 1199 |
+
"step": 760
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.86,
|
| 1203 |
+
"learning_rate": 2.9936823040713464e-07,
|
| 1204 |
+
"logits/chosen": -1.0335241556167603,
|
| 1205 |
+
"logits/rejected": -1.1321046352386475,
|
| 1206 |
+
"logps/chosen": -708.8296508789062,
|
| 1207 |
+
"logps/rejected": -637.4175415039062,
|
| 1208 |
+
"loss": 0.587,
|
| 1209 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1210 |
+
"rewards/chosen": 0.6139867901802063,
|
| 1211 |
+
"rewards/margins": 0.3065374493598938,
|
| 1212 |
+
"rewards/rejected": 0.3074493110179901,
|
| 1213 |
+
"step": 770
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.87,
|
| 1217 |
+
"learning_rate": 2.5486683999940335e-07,
|
| 1218 |
+
"logits/chosen": -1.0177851915359497,
|
| 1219 |
+
"logits/rejected": -1.2519065141677856,
|
| 1220 |
+
"logps/chosen": -767.447021484375,
|
| 1221 |
+
"logps/rejected": -714.9308471679688,
|
| 1222 |
+
"loss": 0.5753,
|
| 1223 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1224 |
+
"rewards/chosen": 0.6363558769226074,
|
| 1225 |
+
"rewards/margins": 0.33481666445732117,
|
| 1226 |
+
"rewards/rejected": 0.30153924226760864,
|
| 1227 |
+
"step": 780
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.88,
|
| 1231 |
+
"learning_rate": 2.137674936387049e-07,
|
| 1232 |
+
"logits/chosen": -1.1295353174209595,
|
| 1233 |
+
"logits/rejected": -1.1564512252807617,
|
| 1234 |
+
"logps/chosen": -883.1470947265625,
|
| 1235 |
+
"logps/rejected": -631.2457275390625,
|
| 1236 |
+
"loss": 0.5517,
|
| 1237 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1238 |
+
"rewards/chosen": 0.695419430732727,
|
| 1239 |
+
"rewards/margins": 0.38906994462013245,
|
| 1240 |
+
"rewards/rejected": 0.30634939670562744,
|
| 1241 |
+
"step": 790
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.89,
|
| 1245 |
+
"learning_rate": 1.7613246905052812e-07,
|
| 1246 |
+
"logits/chosen": -1.0573675632476807,
|
| 1247 |
+
"logits/rejected": -1.1950371265411377,
|
| 1248 |
+
"logps/chosen": -760.9122314453125,
|
| 1249 |
+
"logps/rejected": -630.8699340820312,
|
| 1250 |
+
"loss": 0.5792,
|
| 1251 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1252 |
+
"rewards/chosen": 0.5912384390830994,
|
| 1253 |
+
"rewards/margins": 0.32750827074050903,
|
| 1254 |
+
"rewards/rejected": 0.2637301981449127,
|
| 1255 |
+
"step": 800
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.89,
|
| 1259 |
+
"eval_logits/chosen": -1.0307811498641968,
|
| 1260 |
+
"eval_logits/rejected": -1.1811619997024536,
|
| 1261 |
+
"eval_logps/chosen": -795.125,
|
| 1262 |
+
"eval_logps/rejected": -652.2705078125,
|
| 1263 |
+
"eval_loss": 0.6182109117507935,
|
| 1264 |
+
"eval_rewards/accuracies": 0.6783919334411621,
|
| 1265 |
+
"eval_rewards/chosen": 0.553367018699646,
|
| 1266 |
+
"eval_rewards/margins": 0.2151338756084442,
|
| 1267 |
+
"eval_rewards/rejected": 0.3382331430912018,
|
| 1268 |
+
"eval_runtime": 181.7834,
|
| 1269 |
+
"eval_samples_per_second": 2.189,
|
| 1270 |
+
"eval_steps_per_second": 2.189,
|
| 1271 |
+
"step": 800
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 0.9,
|
| 1275 |
+
"learning_rate": 1.4201879448319356e-07,
|
| 1276 |
+
"logits/chosen": -1.0675430297851562,
|
| 1277 |
+
"logits/rejected": -1.1287426948547363,
|
| 1278 |
+
"logps/chosen": -711.8330078125,
|
| 1279 |
+
"logps/rejected": -551.1827392578125,
|
| 1280 |
+
"loss": 0.5842,
|
| 1281 |
+
"rewards/accuracies": 0.75,
|
| 1282 |
+
"rewards/chosen": 0.5683950185775757,
|
| 1283 |
+
"rewards/margins": 0.2922239601612091,
|
| 1284 |
+
"rewards/rejected": 0.2761710286140442,
|
| 1285 |
+
"step": 810
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 0.91,
|
| 1289 |
+
"learning_rate": 1.1147816229310549e-07,
|
| 1290 |
+
"logits/chosen": -1.016808032989502,
|
| 1291 |
+
"logits/rejected": -1.1210168600082397,
|
| 1292 |
+
"logps/chosen": -691.964599609375,
|
| 1293 |
+
"logps/rejected": -732.0343017578125,
|
| 1294 |
+
"loss": 0.637,
|
| 1295 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1296 |
+
"rewards/chosen": 0.5624631643295288,
|
| 1297 |
+
"rewards/margins": 0.19316788017749786,
|
| 1298 |
+
"rewards/rejected": 0.3692953288555145,
|
| 1299 |
+
"step": 820
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 0.93,
|
| 1303 |
+
"learning_rate": 8.455685061547119e-08,
|
| 1304 |
+
"logits/chosen": -1.1123645305633545,
|
| 1305 |
+
"logits/rejected": -1.132232904434204,
|
| 1306 |
+
"logps/chosen": -859.2161254882812,
|
| 1307 |
+
"logps/rejected": -662.1595458984375,
|
| 1308 |
+
"loss": 0.5684,
|
| 1309 |
+
"rewards/accuracies": 0.75,
|
| 1310 |
+
"rewards/chosen": 0.7145462036132812,
|
| 1311 |
+
"rewards/margins": 0.3542642295360565,
|
| 1312 |
+
"rewards/rejected": 0.36028194427490234,
|
| 1313 |
+
"step": 830
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 0.94,
|
| 1317 |
+
"learning_rate": 6.129565323916814e-08,
|
| 1318 |
+
"logits/chosen": -1.0635792016983032,
|
| 1319 |
+
"logits/rejected": -1.1541740894317627,
|
| 1320 |
+
"logps/chosen": -795.508056640625,
|
| 1321 |
+
"logps/rejected": -634.9236450195312,
|
| 1322 |
+
"loss": 0.5683,
|
| 1323 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1324 |
+
"rewards/chosen": 0.6695567965507507,
|
| 1325 |
+
"rewards/margins": 0.3517037034034729,
|
| 1326 |
+
"rewards/rejected": 0.3178531527519226,
|
| 1327 |
+
"step": 840
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 0.95,
|
| 1331 |
+
"learning_rate": 4.1729817792030004e-08,
|
| 1332 |
+
"logits/chosen": -1.1519498825073242,
|
| 1333 |
+
"logits/rejected": -1.1835066080093384,
|
| 1334 |
+
"logps/chosen": -785.9249267578125,
|
| 1335 |
+
"logps/rejected": -617.5560913085938,
|
| 1336 |
+
"loss": 0.579,
|
| 1337 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1338 |
+
"rewards/chosen": 0.589630126953125,
|
| 1339 |
+
"rewards/margins": 0.2917063534259796,
|
| 1340 |
+
"rewards/rejected": 0.297923743724823,
|
| 1341 |
+
"step": 850
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 0.96,
|
| 1345 |
+
"learning_rate": 2.588899233021358e-08,
|
| 1346 |
+
"logits/chosen": -1.0503435134887695,
|
| 1347 |
+
"logits/rejected": -1.218450665473938,
|
| 1348 |
+
"logps/chosen": -840.5927734375,
|
| 1349 |
+
"logps/rejected": -520.814208984375,
|
| 1350 |
+
"loss": 0.5205,
|
| 1351 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1352 |
+
"rewards/chosen": 0.6971001029014587,
|
| 1353 |
+
"rewards/margins": 0.4537246823310852,
|
| 1354 |
+
"rewards/rejected": 0.24337545037269592,
|
| 1355 |
+
"step": 860
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 0.97,
|
| 1359 |
+
"learning_rate": 1.3797180412583322e-08,
|
| 1360 |
+
"logits/chosen": -1.0805847644805908,
|
| 1361 |
+
"logits/rejected": -1.2713693380355835,
|
| 1362 |
+
"logps/chosen": -800.3570556640625,
|
| 1363 |
+
"logps/rejected": -632.8238525390625,
|
| 1364 |
+
"loss": 0.5758,
|
| 1365 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1366 |
+
"rewards/chosen": 0.5660972595214844,
|
| 1367 |
+
"rewards/margins": 0.31249552965164185,
|
| 1368 |
+
"rewards/rejected": 0.25360172986984253,
|
| 1369 |
+
"step": 870
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 0.98,
|
| 1373 |
+
"learning_rate": 5.4727047281821764e-09,
|
| 1374 |
+
"logits/chosen": -1.0479118824005127,
|
| 1375 |
+
"logits/rejected": -1.3037327527999878,
|
| 1376 |
+
"logps/chosen": -783.6592407226562,
|
| 1377 |
+
"logps/rejected": -529.0799560546875,
|
| 1378 |
+
"loss": 0.5171,
|
| 1379 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1380 |
+
"rewards/chosen": 0.6520611047744751,
|
| 1381 |
+
"rewards/margins": 0.47154098749160767,
|
| 1382 |
+
"rewards/rejected": 0.18052014708518982,
|
| 1383 |
+
"step": 880
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 0.99,
|
| 1387 |
+
"learning_rate": 9.281793319140808e-10,
|
| 1388 |
+
"logits/chosen": -1.0751326084136963,
|
| 1389 |
+
"logits/rejected": -1.2520333528518677,
|
| 1390 |
+
"logps/chosen": -834.0614013671875,
|
| 1391 |
+
"logps/rejected": -488.2327575683594,
|
| 1392 |
+
"loss": 0.5113,
|
| 1393 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1394 |
+
"rewards/chosen": 0.6582953929901123,
|
| 1395 |
+
"rewards/margins": 0.4814949631690979,
|
| 1396 |
+
"rewards/rejected": 0.17680040001869202,
|
| 1397 |
+
"step": 890
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 1.0,
|
| 1401 |
+
"step": 897,
|
| 1402 |
+
"total_flos": 0.0,
|
| 1403 |
+
"train_loss": 0.12323601163483516,
|
| 1404 |
+
"train_runtime": 1012.2753,
|
| 1405 |
+
"train_samples_per_second": 3.544,
|
| 1406 |
+
"train_steps_per_second": 0.886
|
| 1407 |
+
}
|
| 1408 |
+
],
|
| 1409 |
+
"logging_steps": 10,
|
| 1410 |
+
"max_steps": 897,
|
| 1411 |
+
"num_input_tokens_seen": 0,
|
| 1412 |
+
"num_train_epochs": 1,
|
| 1413 |
+
"save_steps": 100,
|
| 1414 |
+
"total_flos": 0.0,
|
| 1415 |
+
"train_batch_size": 1,
|
| 1416 |
+
"trial_name": null,
|
| 1417 |
+
"trial_params": null
|
| 1418 |
+
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4856
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cce9a3a99c6711ebb8b18c4a7e23fb8e32e4da26e087963f99148e32f7380a99
|
| 3 |
size 4856
|