Unverified50% confidenceFactExact time
At the is_prime token, after writing BF16's residual into Q4_K_M, that token's KL dropped from 0.48 after layer 6 to 0.04 after layer 9; the reverse write reproduced the flip from about layer 12, indicating the decisive difference forms between layers 6 and 12.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/17/2026
First Seen
Valid until: 1/15/2027
Sources
Related Entities
Related Claims
UnverifiedIn a comparison of Qwen3-0.6B in BF16 versus Q4_K_M, Q4_K_M changed the top-1 token at 11.6% of positions, with an average KL divergence of 0.067 and perplexity rising from 7.83 to 8.53.77% similarUnverifiedLayer-by-layer denoising experiments between Q4_K_M and BF16 show KL decreasing smoothly from layer 0 to layer 27, with no single layer carrying all the difference.70% similarUnverifiedAcross 12 blocks and 1282 scored positions, when Q4_K_M recovers BF16, attention eliminates 0.030 (95% CI 0.019–0.041) and MLP eliminates 0.036 (0.024–0.048), with the paired difference crossing zero so the two are indistinguishable; when injecting Q4_K_M into BF16, MLP adds more (0.014, 0.005 to 0.023).68% similarUnverifiedIn the scenario where 16 requests each carry a 3,072-token long prompt, Gemma 4 E2B maintains 1049 tokens/second, while Granite 3.3 2B drops to 732 (0.70x), SmolLM3-3B drops to 725 (0.69x), Phi-4-mini drops to 404 (0.39x), and Qwen3-4B reaches only 373 (0.36x).61% similarUnverifiedIn GGUF, BF16 uses 16 bits per parameter, Q8_0 approximately 8 bits, Q4_K_M approximately 4 bits, Q3_K_M approximately 3 bits, and Q2_K approximately 2 bits.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1054505API
curl https://kongchang.com/api/v1/knowledge/claims/1054505MCP
get_claim(id=1054505)