hanzoai/deepspec
2
0
Fork 0
mirror of https://github.com/hanzoai/deepspec.git synced 2026-09-30 21:41:22 +00:00
Code Issues Packages Projects Releases Wiki Activity

release ckpt

Browse Source
This commit is contained in:
Xin Cheng
2026-06-28 20:43:16 +08:00
parent 0a03e19e1c
commit f9879b5c1f
1 changed files with 15 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+15 -2
View File
@@ -48,7 +48,20 @@ bash scripts/eval/eval.sh
`eval.sh` runs `eval.py` against a trained draft checkpoint over the speculative-decoding benchmarks in [eval_datasets/](./eval_datasets/) (gsm8k, math500, aime25, humaneval, mbpp, livecodebench, mt-bench, alpaca, arena-hard-v2). Set: `eval.sh` runs `eval.py` against a trained draft checkpoint over the speculative-decoding benchmarks in [eval_datasets/](./eval_datasets/) (gsm8k, math500, aime25, humaneval, mbpp, livecodebench, mt-bench, alpaca, arena-hard-v2). Set:
- `target_name_or_path` — the target model the draft was trained against (e.g. `Qwen/Qwen3-4B`), - `target_name_or_path` — the target model the draft was trained against (e.g. `Qwen/Qwen3-4B`),
- `draft_name_or_path` — the draft checkpoint, e.g. `~/checkpoints/deepspec/dspark_block8_qwen3_4b/step_latest`. - `draft_name_or_path` — the draft checkpoint, e.g. `~/checkpoints/deepspec/dspark_block8_qwen3_4b/step_latest`, or one of the Hugging Face repo IDs listed in [Released Checkpoints](#released-checkpoints).
### Released Checkpoints
The checkpoints below are the ones used for Table 1 in the paper. They were trained on the `perfect-blend` data generated in non-thinking mode.
| Algorithm | `Qwen/Qwen3-4B` | `Qwen/Qwen3-8B` | `Qwen/Qwen3-14B` | `google/gemma-4-12B-it` |
| --- | --- | --- | --- | --- |
| Eagle3 | [deepseek-ai/eagle3_qwen3_4b_ttt7](https://huggingface.co/deepseek-ai/eagle3_qwen3_4b_ttt7) | [deepseek-ai/eagle3_qwen3_8b_ttt7](https://huggingface.co/deepseek-ai/eagle3_qwen3_8b_ttt7) | [deepseek-ai/eagle3_qwen3_14b_ttt7](https://huggingface.co/deepseek-ai/eagle3_qwen3_14b_ttt7) | [deepseek-ai/eagle3_gemma4_12b_ttt7](https://huggingface.co/deepseek-ai/eagle3_gemma4_12b_ttt7) |
| DFlash | [deepseek-ai/dflash_qwen3_4b_block7](https://huggingface.co/deepseek-ai/dflash_qwen3_4b_block7) | [deepseek-ai/dflash_qwen3_8b_block7](https://huggingface.co/deepseek-ai/dflash_qwen3_8b_block7) | [deepseek-ai/dflash_qwen3_14b_block7](https://huggingface.co/deepseek-ai/dflash_qwen3_14b_block7) | [deepseek-ai/dflash_gemma4_12b_block7](https://huggingface.co/deepseek-ai/dflash_gemma4_12b_block7) |
| DSpark | [deepseek-ai/dspark_qwen3_4b_block7](https://huggingface.co/deepseek-ai/dspark_qwen3_4b_block7) | [deepseek-ai/dspark_qwen3_8b_block7](https://huggingface.co/deepseek-ai/dspark_qwen3_8b_block7) | [deepseek-ai/dspark_qwen3_14b_block7](https://huggingface.co/deepseek-ai/dspark_qwen3_14b_block7) | [deepseek-ai/dspark_gemma4_12b_block7](https://huggingface.co/deepseek-ai/dspark_gemma4_12b_block7) |
> [!IMPORTANT]
> If you cite these results in a new paper, align your setup with the training settings in this repository; otherwise, the comparison is not meaningful. For domain-specific use, fine-tune the draft model again for better results, especially if the target model is expected to run in thinking mode.
## Supported Algorithms ## Supported Algorithms
@@ -68,4 +81,4 @@ DeepSpec builds on the ideas and code of several excellent open-source projects:
- [DFlash](https://github.com/z-lab/dflash) (MIT) — the DFlash draft-model design and training recipe. - [DFlash](https://github.com/z-lab/dflash) (MIT) — the DFlash draft-model design and training recipe.
- [Qwen3](https://github.com/QwenLM/Qwen3) and [Gemma](https://github.com/google-deepmind/gemma) — the target model families supported in this repo. - [Qwen3](https://github.com/QwenLM/Qwen3) and [Gemma](https://github.com/google-deepmind/gemma) — the target model families supported in this repo.
We thank the authors and maintainers of these projects. Contributions of new algorithms are welcome. We thank the authors and maintainers of these projects. Contributions of new algorithms are welcome.