First Divergent Token metric rollouts on GGUF and MLX quants
Parent: Mac local LLMs: Quantization evaluation · Published reference · snapshot 2026-10-05
↓ Facts as markdownall context files
Still true after a fresh search: no FDT measurement on any GGUF or MLX quant of one BF16 base on Apple silicon was found in the search results for "first divergent token quantization GGUF MLX rollout" (results were MLX-vs-GGUF speed posts).
These notes link each claim to its source. A source may be a research report hosted on this site rather than the primary document. A published reference means the content is available; it does not certify independent review or accuracy.Read the editorial policy and follow the sources before relying on a claim.
Facts
- Still true after a fresh search: no FDT measurement on any GGUF or MLX quant of one BF16 base on Apple silicon was found in the search results for "first divergent token quantization GGUF MLX rollout" (results were MLX-vs-GGUF speed posts). [source]
- Already covered by ~/.research/expert/running-llm-models-locally-on-mac/dossiers/branch-and-follow-flip-analysis-harness-for-mac.md; no claim in it is contradicted, and no GGUF or MLX FDT rollout measurement exists in the sources found. [source]
Children
- No children recorded.