Time to update the Elo plot and read angry comments about polynomial regression I guess
@maximelabonne
-
MixEval Benchmark: Evaluating General-Purpose Chatbot Performance
By
–
Next time you pick benchmarks for model evaluation, check this correlation matrix from the MixEval paper. It's a great proxy for @lmsysorg Chatbot Arena performance, perfect for general-purpose chatbots. MixEval: https://
mixeval.github.io -
Daredevil-8B Model Recommendation for Abliteration Issues
By
–
Hahaha thanks for your feedback, I believe this is because of the abliteration process. I would recommend Daredevil-8B if it's an issue for you.
-
Fine-tuning and Model Merging at AI Engineer World’s Fair
By
–
Hey, I'll be speaking at the AI Engineer World's Fair (
@aiDotEngineer
) in SF on June 26. We'll discuss fine-tuning and model merging in concrete terms, with notebooks and examples. -
Wrong Model Switch Boosts Accuracy and Reduces AI Bias
By
–
I realized I was using the wrong model for inference. It's insanely good! Now +62% accuracy, -81% evilness.
-
Kohya_ss Open Source AI Training Tool Repository
By
–
This one https://
github.com/bmaltais/kohya
_ss
… -
Kohya_ss Fine-tuning: Manual UI Process Without Scripts
By
–
No notebook or script unfortunately, it's a lot of clicking with kohya_ss
-
Fine-tuning LLMs with Spectrum: Targeting Key Layers Efficiently
By
–
Fine-tune with Spectrum Dolphin models use this technique to target the most important layers to fine-tune, like in their recent Qwen2-72B version https://
huggingface.co/cognitivecompu
tations/dolphin-2.9.2-qwen2-72b
… You can use their repo to identify the layers and the "unfrozen_parameters" key in your Axolotl config. -
Llama 3 Chat Template Optimization for Fine-tuning
By
–
Thanks actually I realised that Axolotl had a Llama 3 chat template after this finetune. It'd be better to use it instead of chatml for sure. It would be better but tbh I've milked this recipe a bit too much 🙂