Fine Tuning

concepts · 14 notes linked

Related: Large Language Models · Chatgpt · Openai · GPT-4 · Google · Hugging Face · LLM Evaluation · Langchain

Notes

• Fine-tuning ChatGPT: Surpassing GPT-4 Summarization Performance — 63% Cost Reduction and 11x Speed Enhancement using Synthetic Data and LangSmith — Fine-tuned ChatGPT beats GPT-4 summarization at 63% lower cost and 11x faster using chain-of-density synthetic data
• Fine-tuning · Hugging Face — Fine-tuning pretrained LLMs with Hugging Face Trainer API
• Gemma: Introducing new state-of-the-art open models — Google DeepMind's Gemma open-weight model family launch announcement
• How to fine-tune FunctionGemma and run it locally — Fine-tuning Google's 270M FunctionGemma locally with Unsloth
• IBM, HuggingFace, and NASA Open-Sources Watsonx.ai Foundation Model — Open-source geospatial foundation model for climate and Earth research
• Learning to Replicate Expert Judgment in Financial Tasks — Fine-tuned LLM beats frontier models at financial info triage
• Must-Have Prompt Engineering Skills for 2024 — In-demand prompt engineering skills, platforms, and tools
• RAFT: Adapting Language Model to Domain Specific RAG — Fine-tuning recipe training LLMs to ignore distractor documents in domain RAG
• Researchers at Boston University Release the Platypus Family of Fine-Tuned LLMs — Cheap fast LLM fine-tuning via curated Open-Platypus dataset and LoRA merging
• The False Promise of Imitating Proprietary LLMs — Finetuning open models on ChatGPT outputs mimics style but not factuality or capability
• Tx-LLM: Supporting therapeutic development with large language models — PaLM-2 fine-tuned on 66 drug discovery tasks across full pipeline
• Unbabel says its new AI model has dethroned OpenAI's GPT-4 as the tech industry's best language translator — TowerLLM 7B/13B model edges GPT-4o on multilingual translation benchmarks
• Vicuna: An Open-Source Chatbot Impressing GPT-4 with 90%* ChatGPT Quality — LLaMA fine-tune on ShareGPT data achieving near-ChatGPT quality for $300
• We Got Claude to Fine-Tune an Open Source LLM — Hugging Face skill enables Claude Code to submit and manage LLM training jobs