No official torrent yet — requesting it tells the flock what to host next. You cannot seed a catalog-only page.
Cataloged from ISTA-DASLab/Qwen3.5-4B-GGUF-GSQ
All models by ISTA-DASLab →GGUF Q2_K_XL checkpoint of Qwen/Qwen3.5-4B
in which the discrete grid assignments have been refined with GSQ
(Gumbel-Softmax Quantization), starting from the public Unsloth GGUF
initialization and projected back into the same K-Quant format.
The optimized file runs unchanged on llama.cpp / Ollama and is a
drop-in replacement for the corresponding standard GGUF checkpoint.
Qwen/Qwen3.5-4BUD-Q2_K_XLQ2_K_XLllama.cpp, ollama, LM Studio, anything that consumes GGUFThis file is bit-for-bit standard GGUF Q2_K. GSQ only changes the values of the quantized weights by relearning the discrete grid assignments inside each K-Quant block.
As a result:
llama.cpp / ollama build that loads a regular Qwen3.5-4B Q2_K_XL GGUF
should load this file with zero changes.Q2_K_XL, rather than
per-tensor dtypes. This refers to the on-disk K-Quant encoding, not the
precision of any optimizer state used during GSQ refinement.We compare the GSQ-refined checkpoint against the Unsloth UD-Q2_K_XL
checkpoint used as initialization.
All evaluations are zeroshot. The Unsloth UD-Q2_K_XL model was evaluated with thinking disabled on all
tasks, because its thinking mode was found to be broken and produced worse
results.
For the GSQ-refined model:
For MMLU-Pro, both models were evaluated with a maximum output budget of 32K tokens.
hf download ISTA-DASLab/Qwen3.5-4B-GGUF-GSQ \ Qwen3.5-4B-Q2_K_XL.gguf --local-dir .
./llama-cli -m Qwen3.5-4B-Q2_K_XL.gguf -p "Hello"