22 lines
1005 B
Markdown
22 lines
1005 B
Markdown
---
|
|
license: apache-2.0
|
|
datasets:
|
|
- anon8231489123/ShareGPT_Vicuna_unfiltered
|
|
language:
|
|
- en
|
|
pipeline_tag: text-generation
|
|
---
|
|
## Model description
|
|
This is a Vicuna-like model with only 68M parameters, which is fine-tuned from [LLaMA-68m](https://huggingface.co/JackFram/llama-68m) on [ShareGPT](https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered) data.
|
|
|
|
The training setup follows the [Vicuna suite](https://github.com/lm-sys/FastChat).
|
|
|
|
The model is mainly developed as a base Small Speculative Model in the [MCSD paper](https://arxiv.org/pdf/2401.06706.pdf). As a comparison, it can be better aligned to the Vicuna models than LLaMA-68m with little loss of alignment to the LLaMA models.
|
|
|
|
|
|
| Draft Model | Target Model | Alignment |
|
|
| -------------- | ------------- | --------- |
|
|
| LLaMA-68/160M | LLaMA-13/33B | 😃 |
|
|
| LLaMA-68/160M | Vicuna-13/33B | 😟 |
|
|
| Vicuna-68/160M | LLaMA-13/33B | 😃 |
|
|
| Vicuna-68/160M | Vicuna-13/33B | 😃 | |