|
Download README.md from sebaiinc/Seb-AI-S1-Mini-409M: direct link, hf CLI and curl.
- Browser
- Download file 1.51 kB
-
https://huggingface.co/sebaiinc/Seb-AI-S1-Mini-409M/resolve/main/README.md
- Command line
-
hf download hf://sebaiinc/Seb-AI-S1-Mini-409M/README.md
-
curl -L -o README.md https://huggingface.co/sebaiinc/Seb-AI-S1-Mini-409M/resolve/main/README.md
1.51 kB
| license: apache-2.0 | |
| language: | |
| - en | |
| tags: | |
| - seb-ai | |
| - text-generation | |
| - causal-lm | |
| - trained-from-scratch | |
| pipeline_tag: text-generation | |
| # Seb AI S1 Mini (350M) | |
| **Seb AI S1 Mini** is an open-source language model trained completely from scratch by **Seb AI Inc.** | |
| - **Official Website:** [https://sebaioffical.lovable.app](https://sebaioffical.lovable.app) | |
| - **Official API:** `https://sebaioffical.lovable.app/api/sebaiapi` | |
| - **Creator:** Seb AI Inc. | |
| - **Licence:** Apache 2.0 (Open-source for anyone to download, use, and fine-tune) | |
| ## Model Details | |
| - **Parameters:** ~409 Million (350M-class architecture) | |
| - **Architecture:** Llama-style Causal Language Model (`LlamaForCausalLM`) | |
| - **Context Length:** 512 tokens | |
| - **Training:** Trained from scratch (random initialization) on educational Q&A instruction data. | |
| ## Prompt Format | |
| To chat with **Seb AI S1 Mini** or fine-tune it on your own dataset, use this exact format: | |
| ```text | |
| User: What is artificial intelligence? | |
| Assistant: | |
| ``` | |
| ## Quickstart Python Code | |
| ```python | |
| from transformers import pipeline | |
| pipe = pipeline("text-generation", model="Sebestianmek/Seb-AI-S1-Mini-350M") | |
| output = pipe("User: Hello! Who made you?\nAssistant:", max_new_tokens=100, do_sample=True, temperature=0.7) | |
| print(output["generated_text"]) | |
| ``` | |
| ## Fine-Tuning | |
| Because **Seb AI S1 Mini** uses the standard Llama architecture and is released under Apache 2.0, you can easily fine-tune it using **Unsloth**, **LLaMA-Factory**, **Axolotl**, or Hugging Face `TRL`. |