DATAandClaude Opus 5 09230bc499 Measure --reasoning-budget: a real cap, unlike --predict ...

Counted the thinking block exactly via /tokenize. Server default stops it at
4095 tokens; a client sending reasoning_budget=16384 still gets 4095, so it
cannot be raised per request. Changing it means editing the compose file.

It truncates mid-derivation on a moderately hard problem, but the model
recovered and answered correctly (empty solution set, verified by brute
force), so there is no evidence 4096 is actually damaging output.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

2026-07-25 15:03:49 +02:00

..

bench.py

llama.cpp CUDA runner for Qwen3.6-35B-A3B at 128k context

2026-07-25 13:49:33 +02:00

reasoning_budget_probe.py

Measure --reasoning-budget: a real cap, unlike --predict

2026-07-25 15:03:49 +02:00

spec_bench.py

Measure speculative decoding: all ngram modes lose to plain decoding

2026-07-25 13:54:08 +02:00

sweep.sh

llama.cpp CUDA runner for Qwen3.6-35B-A3B at 128k context

2026-07-25 13:49:33 +02:00