初始化项目，由ModelHub XC社区提供模型

Model: W-61/llama-3-8b-base-new-dpo-hh-harmless-s_star0.6-4xh200-batch-64-20260421-213851 Source: Original Platform
2026-06-01 08:07:27 +08:00
commit 765dad5674
682 changed files with 16796 additions and 0 deletions
--- a/README.md
+++ b/README.md
@@ -0,0 +1,81 @@
+---
+library_name: transformers
+base_model: W-61/llama-3-8b-base-sft-hh-harmless-4xh200
+tags:
+- alignment-handbook
+- new-dpo
+- generated_from_trainer
+datasets:
+- Anthropic/hh-rlhf
+model-index:
+- name: llama-3-8b-base-new-dpo-hh-harmless-s_star0.6-4xh200-batch-64-20260421-213851
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# llama-3-8b-base-new-dpo-hh-harmless-s_star0.6-4xh200-batch-64-20260421-213851
+
+This model is a fine-tuned version of [W-61/llama-3-8b-base-sft-hh-harmless-4xh200](https://huggingface.co/W-61/llama-3-8b-base-sft-hh-harmless-4xh200) on the Anthropic/hh-rlhf dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.5427
+- Fcm Dpo/beta: 0.0127
+- Fcm Dpo/q T: 0.3797
+- Fcm Dpo/delta: -0.0085
+- Fcm Dpo/margin: 46.9651
+- Margin Dpo/margin Mean: 46.9651
+- Margin Dpo/margin Std: 76.4295
+- Logps/chosen: -176.9747
+- Logps/rejected: -228.6294
+- Logps/ref Chosen: -74.8595
+- Logps/ref Rejected: -79.5490
+- Logits/chosen: 0.6471
+- Logits/rejected: 0.5976
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 5e-07
+- train_batch_size: 8
+- eval_batch_size: 8
+- seed: 42
+- distributed_type: multi-GPU
+- num_devices: 4
+- gradient_accumulation_steps: 2
+- total_train_batch_size: 64
+- total_eval_batch_size: 32
+- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
+- lr_scheduler_type: cosine
+- lr_scheduler_warmup_ratio: 0.1
+- num_epochs: 1
+
+### Training results
+
+| Training Loss | Epoch  | Step | Validation Loss | Fcm Dpo/beta | Fcm Dpo/q T | Fcm Dpo/delta | Fcm Dpo/margin | Margin Dpo/margin Mean | Margin Dpo/margin Std | Logps/chosen | Logps/rejected | Logps/ref Chosen | Logps/ref Rejected | Logits/chosen | Logits/rejected |
+|:-------------:|:------:|:----:|:---------------:|:------------:|:-----------:|:-------------:|:--------------:|:----------------------:|:---------------------:|:------------:|:--------------:|:----------------:|:------------------:|:-------------:|:---------------:|
+| 1.0309        | 0.3023 | 200  | 0.5479          | 0.0706       | 0.3810      | -0.0076       | 8.4404         | 8.4404                 | 14.1426               | -88.2471     | -101.3770      | -74.8595         | -79.5490           | 0.3430        | 0.2953          |
+| 1.058         | 0.6047 | 400  | 0.5442          | 0.0156       | 0.3824      | 0.0052        | 37.3081        | 37.3081                | 61.1427               | -151.6879    | -193.6855      | -74.8595         | -79.5490           | 0.6412        | 0.5884          |
+| 1.1526        | 0.9070 | 600  | 0.5427          | 0.0127       | 0.3797      | -0.0085       | 46.9651        | 46.9651                | 76.4295               | -176.9747    | -228.6294      | -74.8595         | -79.5490           | 0.6471        | 0.5976          |
+
+
+### Framework versions
+
+- Transformers 4.51.0
+- Pytorch 2.3.1+cu121
+- Datasets 2.21.0
+- Tokenizers 0.21.4