Downloads · 30 days
62
11% of all-time downloads
MistralThinker-v1.1-Reasoner-256-gguf
sandmanbuzz/MistralThinker-v1.1-Reasoner-256-gguf
MistralThinker-v1.1-Reasoner-256-gguf is a machine learning model from sandmanbuzz. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as other.
q4km q5km q6k and the fp16 if you want to make more
All-time downloads
561
Public
Repo size
96.8 GB
Likes
1
Public
Repo makeup
Click a slice to open those files.
.gguf96.8 GB · 100%
At a glance
- License
- other
- Access
- Public
- Created
- Apr 14, 2025
- Updated
- Apr 16, 2025
- SHA
- 59f61d80
Base models
Model card
From the Hugging Face model README
Quants
as of lcpp d6d2c2ab8c8865784ba9fef37f2b2de3f2134d33
- q4_k_m
- q5_k_m
- q6_k
- and the fp16 if you want to make more
Undi95/MistralThinker-v1.1 but with DavidAU/Mistral-Small-3-Reasoner-s1-24B-LORA-256-RANK slapped over it.
It seems pretty nice. Here's whatever settings I happened to have sitting in ST when I tried it:
t=1.25
topk=45
rep.pen=1.03
rep.pen.range=384
rep.pen.slope=0.7
smoothing=1.7
Personally I liked the output more than rawdog Thinker, which felt kinda model-y in terms of phrasing. Having said that, it seems a little less instruct-y than the pure thing. That could very well be because I literally randomly picked which LORA to stick on it. Zero science or care went into the product, here, it's just smashing rocks together for no reason. It's very likely I made the moddel crappier as an instruct as a result of being so ham-fisted.
- License
- other
- Created
- Apr 14, 2025
- Updated
- Apr 16, 2025