README.md

---
license: apache-2.0
datasets:
- HuggingFaceTB/smollm-corpus
base_model:
- HuggingFaceTB/SmolLM-1.7B
pipeline_tag: text-generation
---

**Research Paper** ["Towards Economical Inference: Enabling DeepSeek's Multi-Head Latent Attention in Any Transformer-based LLMs"](https://arxiv.org/abs/2502.14837)

## Inference

- Step 1: Download the [**monkey patch file**](https://github.com/JT-Ushio/MHA2MLA/blob/main/src/mha2mla/monkey_patch.py).
```shell
wget https://raw.githubusercontent.com/JT-Ushio/MHA2MLA/refs/heads/main/src/mha2mla/monkey_patch.py
```

- Step 2(Option): For MHA2MLA models using Partial-RoPE 2-nrom method, Download the [**qk_2-norm file**](https://github.com/JT-Ushio/MHA2MLA/tree/main/utils). 
Take `qk_tensor_1.7B.pth` as an example:
```shell
wget https://github.com/JT-Ushio/MHA2MLA/raw/refs/heads/main/utils/qk_tensor_1.7B.pth
```

- Step 3: Download the [MHA2MLA models](https://huggingface.co/fnlp/SmolLM-1B7-MLA-d_kv_16) and run inference. 
Take `fnlp/SmolLM-1B7-MLA-d_kv_16` as an example:

```python
import torch
from transformers import AutoConfig, AutoTokenizer, LlamaForCausalLM
from monkey_patch import infer_monkey_patch

model_name = "fnlp/SmolLM-1B7-MLA-d_kv_16"

# Monkey Patch: MHA -> MLA
config = AutoConfig.from_pretrained(model_name)
if "RoPE" in config:
    config.RoPE["qk_tensor_path"] = "qk_tensor_1.7B.pth"  # Configuration for Specific Models
    infer_monkey_patch(config.RoPE)

tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
model = LlamaForCausalLM.from_pretrained(model_name, config=config, torch_dtype=torch.bfloat16).cuda()

# Generate
text = "Which American-born Sinclair won the Nobel Prize for Literature in 1930?"
inputs = tokenizer(text, return_tensors="pt").to(model.device)
generation_kwargs = {"do_sample": False, "use_cache": True, "max_new_tokens": 128}
output = model.generate(**inputs, **generation_kwargs)

print(tokenizer.decode(output[0], skip_special_tokens=True))
# - Sinclair Lewis
```

## Citation
```
@misc{ji2025economicalinferenceenablingdeepseeks,
      title={Towards Economical Inference: Enabling DeepSeek's Multi-Head Latent Attention in Any Transformer-based LLMs}, 
      author={Tao Ji and Bin Guo and Yuanbin Wu and Qipeng Guo and Lixing Shen and Zhan Chen and Xipeng Qiu and Qi Zhang and Tao Gui},
      year={2025},
      eprint={2502.14837},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2502.14837}, 
}
```