| language: en | |
| license: apache-2.0 | |
| datasets: | |
| - bexgboost/openai-agents-python-qa-firecrawl | |
| tags: | |
| - firecrawl | |
| - gemma | |
| - web-scraping | |
| - question-answering | |
| - fine-tuned | |
| - unsloth | |
| # Gemma-3-12B Firecrawl Expert | |
| This is a fine-tuned version of Gemma-3-12B specialized in answering questions about Firecrawl web scraping. | |
| ## Model Details | |
| - **Base Model:** unsloth/gemma-3-12b-it | |
| - **Fine-tuning Method:** LoRA (Low-Rank Adaptation) using Unsloth | |
| - **Training Library:** Unsloth | |
| - **Fine-tuned:** April 24, 2025 | |
| ## Training Data | |
| The model was fine-tuned on the bexgboost/openai-agents-python-qa-firecrawl dataset, which contains question-answer pairs about Firecrawl and web scraping techniques. | |
| ## Use Cases | |
| This model is specialized in: | |
| - Answering questions about Firecrawl web scraping | |
| - Providing guidance on web scraping techniques | |
| - Helping with Firecrawl implementation | |
| ## Training Parameters | |
| - **LoRA Rank:** 8 | |
| - **LoRA Alpha:** 8 | |
| - **Learning Rate:** 2e-4 | |
| - **Epochs:** 1 | |
| - **Quantization:** 4-bit | |
| ## Usage with Unsloth | |
| ```python | |
| from unsloth import FastModel | |
| model, tokenizer = FastModel.from_pretrained( | |
| model_name = "Laksh99/Gemma_finetuned_april_24_2025", | |
| max_seq_length = 2048, | |
| load_in_4bit = True | |
| ) | |
| ``` | |