| # Repro - Float8@2bits: Entropy Coding Enables Data-Free Model Compression | |
| ## Pages | |
| | Page | | |
| | --- | | |
| | [Claim 1: compression time and GPU ANS](#/claim-1-compression-time-and-gpu-ans) | | |
| | [Claim 2: LLaMA-2 7B C4 perplexity](#/claim-2-llama-2-7b-c4-perplexity) | | |
| | [Claim 3: calibrated-method comparison](#/claim-3-calibrated-method-comparison) | | |
| | [Claim 4: representational expressivity](#/claim-4-representational-expressivity) | | |
| | [Claim 5: inference throughput](#/claim-5-inference-throughput) | | |
| | [Claim 6: instruction-tuned benchmarks](#/claim-6-instruction-tuned-benchmarks) | | |
| | [Conclusion](#/conclusion) | | |
| | [float8-2bits-repro](#/float8-2bits-repro) | | |
| | [Claim 1: LLPF_M2M low-loss connectivity](#/claim-1-llpf-m2m-low-loss-connectivity) | | |
| | [Claim 2: Layer-wise distance consistency](#/claim-2-layer-wise-distance-consistency) | | |
| | [Claim 3: LLPF versus AutoNEB](#/claim-3-llpf-versus-autoneb) | | |
| | [Claim 4: LLPF_M2O toward origin](#/claim-4-llpf-m2o-toward-origin) | | |