Search results

Text-only PARO MoE checkpoints ship a vision_config → servers mis-route to VLM engine (load failure / SIGKILL)

Status: Open.

#49 In z-lab/paroquant;

· sangemaru opened on May 22, 2026

Qwen3.6-35B-A3B-PARO: config.json declares mtp_num_hidden_layers=1 but no MTP weights shipped → 0% accept + ~50% decode overhead by default

Status: Open.

#47 In z-lab/paroquant;

· sangemaru opened on May 19, 2026

Gemma-4 26B-A4B PARO: Metal GPU command-buffer OOM during MoE forward pass on M1 Max (not VLM-path SIGKILL)

Status: Open.

#46 In z-lab/paroquant;

· sangemaru opened on May 19, 2026

when will we upload qwen3-32b-paro model in huggingface?

Status: Open.

#45 In z-lab/paroquant;

· dongdongvs6 opened on May 18, 2026

Request Jackrong/Qwopus3.5-27B-v3 Paro version

model requestRequest a new model to be quantizedRequest a new model to be quantized

Status: Open.

#27 In z-lab/paroquant;

· menciuschow opened on Apr 5, 2026

rocm support

Status: Open.

#25 In z-lab/paroquant;

· ambyjkl opened on Mar 29, 2026

Can't disable thinking/reasoning mode + JSON format issues with OpenAI-compatible API

Status: Open.

#24 In z-lab/paroquant;

· Milor123 opened on Mar 23, 2026

Model Requests

model requestRequest a new model to be quantizedRequest a new model to be quantized

Status: Open.

#23 In z-lab/paroquant;

· copilotprosub1 opened on Mar 22, 2026

Request for 2-bit, 3-bit, and 4-bit quantized versions of LLaMA 3.1-70B and LLaMA 2-70B

model requestRequest a new model to be quantizedRequest a new model to be quantized

Status: Open.

#22 In z-lab/paroquant;

· Hittopu opened on Mar 20, 2026

Empty models list returned from /v1/models endpoint

Status: Open.

#20 In z-lab/paroquant;

· zlatent opened on Mar 19, 2026

can you have a benchmark on unsloth UD2.0 with the same quant bit

Status: Open.

#13 In z-lab/paroquant;

· cresseelia opened on Mar 11, 2026