--- base_model: LiquidAI/LFM2.5-350M library_name: transformers tags: - text-generation - text-editing - rewriting - paraphrasing - instruct private: true license: other language: - en pipeline_tag: text-generation ---

cubby

In this repository, we propose the next iteration of `palmer`, a new family of small language models trained with better foundational models, better data but same tasks: Text rewriting, paraphrasing, tone transfer, grammar-style editing, the important distinction is *instruction-following text transformations*. With only 350m parameters, this model is perfect for experiments on laptops and mobile devices. #### prompt There is no prompt intentionally set. #### Leaderboard > Our custom evaluation consists of 100 text-editing tasks where the model has to modify a diversity of texts in different ways. All models were evaluated using `q8_0` gguf quantization. | Model | Avg | Emb | String | Token | Rule | Exact | Strict | Flagged | Params | |---:|---|---:|---:|---:|---:|---:|---:|---:|---| | `🥇 core` | **88.53** | 96.56 | 85.19 | 82.33 | 91.14 | 51/100 | 75/100 | 10/100 | 350m | | `nano` | 84.70 | 95.81 | 81.12 | 78.22 | 87.57 | 39/100 | 68/100 | 19/100 | 90m | | `lfm2` | 72.04 | 93.19 | 62.00 | 56.22 | 75.35 | 21/100 | 40/100 | 35/100 | 700m | As people already know, leaderboards are not enough, they're just a small light in the right direction. So, if you find its weaknesses, don't hesitate to share them, this way it's ensured a better experience for the whole community. #### supporters Buy Me A Coffee **Note**: the model has not been tested as a chat assistant and it might not work as intended, use with caution.