I went and checked. The loops on this one aren't it debating whether to answer. That language fades as the trace goes on and it just keeps rewriting the answer inside its thinking. When it stops, the answer's complete. Its a stopping problem, not hidden refusal. Only looked at this model though.
Josh Bolding PRO
BoldingBuilds
P(doom) 1%
AI & ML interests
Independent builder working on local AI, reasoning models, and practical agents. I test models, experiment with small weight edits, and share results, GGUFs, and evaluation tools.
Recent Activity
new activity about 4 hours ago
BoldingBuilds/Qwen3.8-27B-Abliterated-ThinkFix-GGUF:great release !!! published a model about 4 hours ago
BoldingBuilds/AREX-2-Abliterated-GGUF updated a collection about 4 hours ago
AREX-2Organizations
replied to their post 1 day ago
posted an update 1 day ago
Post
4468
Abliterated thinking models have a habit: they decide what to say, then keep arguing with themselves until the token budget runs out. Qwen3.8-27B Abliterated ThinkFix is an uncensored Qwen3.8-27B with one output row edited so the model closes its thinking when it's ready to answer.
On 40 sensitive prompts the edit never saw, 8k output cap: 33 replies finished cleanly instead of 22, 5 hit the limit instead of 17, 61 minutes for the set instead of 75.
Agent use was checked four ways and the edit changes nothing there. MTP draft head kept. Q4_K_M to Q8_0, each tested after the edit. The card lists what the edit was fit on and what it does not do.
BoldingBuilds/Qwen3.8-27B-Abliterated-ThinkFix-GGUF
On 40 sensitive prompts the edit never saw, 8k output cap: 33 replies finished cleanly instead of 22, 5 hit the limit instead of 17, 61 minutes for the set instead of 75.
Agent use was checked four ways and the edit changes nothing there. MTP draft head kept. Q4_K_M to Q8_0, each tested after the edit. The card lists what the edit was fit on and what it does not do.
BoldingBuilds/Qwen3.8-27B-Abliterated-ThinkFix-GGUF
- 5 replies
posted an update 7 days ago
Post
235
I released a one-row weight edit for Qwen3.8 27B and Flash-Next to address a frustrating failure: using the entire output budget thinking, then returning no final answer.
The edit changes only the output-layer row that scores </think> and is packaged into ordinary GGUF files.
On 200 MATH-500 problems at a 4,096-token output limit:
• 27B: 31 empty answers → 0
• Flash-Next: 29 → 0
For comparison, llama.cpp's --reasoning-budget 2048 also eliminated blanks, with similar accuracy. The aim here is to put the control in the model file.
The report includes correctness scores, regressions, compatibility checks, evaluation limitations, and links to both models.
Report and downloads:
BoldingBuilds/helping-qwen-finish-thinking
Built on Qwen's models and Unsloth's GGUF conversions. If you've worked with local reasoning models, I'd like to hear what I got wrong.
The edit changes only the output-layer row that scores </think> and is packaged into ordinary GGUF files.
On 200 MATH-500 problems at a 4,096-token output limit:
• 27B: 31 empty answers → 0
• Flash-Next: 29 → 0
For comparison, llama.cpp's --reasoning-budget 2048 also eliminated blanks, with similar accuracy. The aim here is to put the control in the model file.
The report includes correctness scores, regressions, compatibility checks, evaluation limitations, and links to both models.
Report and downloads:
BoldingBuilds/helping-qwen-finish-thinking
Built on Qwen's models and Unsloth's GGUF conversions. If you've worked with local reasoning models, I'd like to hear what I got wrong.
posted an update 12 days ago
Post
360
I tested every uncensored Ternary Bonsai 2 27B on the Hub, all 11 builds from 6 uploaders, including my own. Same prompts, same judge, same GPUs, every file pinned by sha256.
• Best answers: @Hikari07jp and @dealignai (~0.94 answer quality with thinking off)
• Only edit with no measurable MMLU cost: mine (±0.15 pp). Every other edit loses 0.64–2.68 pp, all p < 0.001
• Heretic and Blackfrost still refuse 11–12% of harmful prompts
• Thinking mode at 4,096 tokens: 5–31% of harmful prompts get no answer. PrismML recommends 16,384+, and I'm rerunning at that budget
Full report, charts and model-card checks: BoldingBuilds/bonsai-2-uncensored-shootout
• Best answers: @Hikari07jp and @dealignai (~0.94 answer quality with thinking off)
• Only edit with no measurable MMLU cost: mine (±0.15 pp). Every other edit loses 0.64–2.68 pp, all p < 0.001
• Heretic and Blackfrost still refuse 11–12% of harmful prompts
• Thinking mode at 4,096 tokens: 5–31% of harmful prompts get no answer. PrismML recommends 16,384+, and I'm rerunning at that budget
Full report, charts and model-card checks: BoldingBuilds/bonsai-2-uncensored-shootout