Hugging Face's logo Hugging Face
  • Models
  • Datasets
  • Spaces
  • Buckets new
  • Docs
  • Enterprise
  • Pricing
    • Website
      • Tasks
      • HuggingChat
      • Collections
      • Languages
      • Organizations
    • Community
      • Blog
      • Posts
      • Daily Papers
      • Hardware
      • Learn
      • Discord
      • Forum
      • GitHub
    • Solutions
      • Team & Enterprise
      • Hugging Face PRO
      • Enterprise Support
      • Inference Providers
      • Inference Endpoints
      • Storage Buckets

  • Log In
  • Sign Up

Spaces:
vimarsh
/
repro-float8-2bits-entropy-coding-enables-data-free-model-compression
Running

App Files Files Community
repro-float8-2bits-entropy-coding-enables-data-free-model-compression / pages /index.md
vimarsh's picture
vimarsh
Update logbook: Repro - Float8@2bits: Entropy Coding Enables Data-Free Model Compression
43eff34 verified 13 days ago
preview code
|
Raw
History Blame Contribute Delete
979 Bytes

Repro - Float8@2bits: Entropy Coding Enables Data-Free Model Compression

Pages

Page
Claim 1: compression time and GPU ANS
Claim 2: LLaMA-2 7B C4 perplexity
Claim 3: calibrated-method comparison
Claim 4: representational expressivity
Claim 5: inference throughput
Claim 6: instruction-tuned benchmarks
Conclusion
float8-2bits-repro
Claim 1: LLPF_M2M low-loss connectivity
Claim 2: Layer-wise distance consistency
Claim 3: LLPF versus AutoNEB
Claim 4: LLPF_M2O toward origin