Felix Fischer's picture

Felix Fischer

FlipTip

·

AI & ML interests

None yet

Recent Activity

upvoted an article 8 days ago

Qwen3.5: Nobody Agrees on Attention Anymore

liked a model 4 months ago

maya-research/maya1

reacted to codelion's post with 🔥 6 months ago

I wanted to share a technique that's been working really well for recovering performance after INT4 quantization. Typically, quantizing the LLM to INT4 (unlike say INT8) for inference can incur some accuracy loss. Instead of accepting the quality loss, we used the FP16 model as a teacher to train a tiny LoRA adapter (rank=16) for the quantized model. The cool part: the model generates its own training data using the Magpie technique so no external datasets needed. This is critical because we want to remain as much as possible in the distribution of the model's natural responses. Last year Apple's foundational models paper (https://arxiv.org/pdf/2407.21075) had proposed a similar technique and found "By using accuracy-recovery LoRA adapters with only rank 16, Alpaca win rate can be improved by 7-18%, GMS8K accuracy is boosted by 5-10%." (page 47). We saw similar results on Qwen3-0.6B: Perplexity: 2.40 → 2.09 (only 5.7% degradation from FP16 baseline) Memory: Only 0.28GB vs 1.0GB for FP16 (75% reduction) Speed: 3.0x faster inference than FP16 Quality: Generates correct, optimized code solutions - Pre-trained adapter: https://huggingface.co/codelion/Qwen3-0.6B-accuracy-recovery-lora - GitHub repo: https://github.com/codelion/ellora Happy to answer questions about the implementation or help anyone trying to replicate this. The key insight is that quantization errors are systematic and learnable - a small adapter can bridge the gap without negating the benefits of quantization. Has anyone else experimented with self-distillation for quantization recovery? Would love to hear about different approaches!

View all activity

Organizations

None yet

FlipTip 's collections 7

FlipTip (Felix Fischer)

Felix Fischer's picture

Felix Fischer

FlipTip

·

AI & ML interests

None yet

Recent Activity

upvoted an article 8 days ago

Qwen3.5: Nobody Agrees on Attention Anymore

liked a model 4 months ago

maya-research/maya1

reacted to codelion's post with 🔥 6 months ago

I wanted to share a technique that's been working really well for recovering performance after INT4 quantization. Typically, quantizing the LLM to INT4 (unlike say INT8) for inference can incur some accuracy loss. Instead of accepting the quality loss, we used the FP16 model as a teacher to train a tiny LoRA adapter (rank=16) for the quantized model. The cool part: the model generates its own training data using the Magpie technique so no external datasets needed. This is critical because we want to remain as much as possible in the distribution of the model's natural responses. Last year Apple's foundational models paper (https://arxiv.org/pdf/2407.21075) had proposed a similar technique and found "By using accuracy-recovery LoRA adapters with only rank 16, Alpaca win rate can be improved by 7-18%, GMS8K accuracy is boosted by 5-10%." (page 47). We saw similar results on Qwen3-0.6B: Perplexity: 2.40 → 2.09 (only 5.7% degradation from FP16 baseline) Memory: Only 0.28GB vs 1.0GB for FP16 (75% reduction) Speed: 3.0x faster inference than FP16 Quality: Generates correct, optimized code solutions - Pre-trained adapter: https://huggingface.co/codelion/Qwen3-0.6B-accuracy-recovery-lora - GitHub repo: https://github.com/codelion/ellora Happy to answer questions about the implementation or help anyone trying to replicate this. The key insight is that quantization errors are systematic and learnable - a small adapter can bridge the gap without negating the benefits of quantization. Has anyone else experimented with self-distillation for quantization recovery? Would love to hear about different approaches!

View all activity

Organizations

None yet

FlipTip 's collections 7