| language: | |
| - en | |
| license: other | |
| library_name: pytorch | |
| tags: | |
| - causal-lm | |
| - chat | |
| - chatml | |
| - instruction-tuned | |
| - safetensors | |
| - small-language-model | |
| - meet25m | |
| # Meet25M ChatML | |
| ChatML fine-tuned version of Meet25M Base. | |
| ## Model | |
| - Base: Meet25M Base | |
| - Approx size: ~25M parameters | |
| - Context length: 1024 | |
| - Tokenizer: custom byte-level BPE with ChatML tokens | |
| - Chat format: ChatML | |
| ## ChatML Format | |
| Use: | |
| ```text | |
| <|im_start|>system | |
| You are a helpful assistant.<|im_end|> | |
| <|im_start|>user | |
| Your message here.<|im_end|> | |
| <|im_start|>assistant | |
| ``` | |
| The model should generate the assistant response until `<|im_end|>`. | |
| ## SFT Dataset | |
| Fine-tuned on: | |
| - `OpenLeecher/lmsys_chat_1m_clean` | |
| ## Files | |
| - `model.safetensors` — safetensors checkpoint | |
| - `config.json` — model config | |
| - `tokenizer/` — tokenizer files | |
| - `safetensors_info.json` — checkpoint metadata | |
| ## Loading | |
| This is not a standard Transformers `AutoModelForCausalLM` checkpoint. | |
| Use the custom GPT class from the training/SFT script and load `model.safetensors`. | |