Hugging Face
Models
Datasets
Spaces
Buckets
new
Docs
Enterprise
Pricing
Website
Tasks
HuggingChat
Collections
Languages
Organizations
Community
Blog
Posts
Daily Papers
Learn
Discord
Forum
GitHub
Solutions
Team & Enterprise
Hugging Face PRO
Enterprise Support
Inference Providers
Inference Endpoints
Storage Buckets
Log In
Sign Up
rdtand
/
Qwen3.6-35B-A3B-PrismaQuant-4.75bit-vllm
like
22
Image-Text-to-Text
Safetensors
English
multilingual
vllm
qwen3_5_moe
qwen3.6
Mixture of Experts
vision-language
multimodal
deltanet
quantized
mixed-precision
nvfp4
mxfp8
compressed-tensors
prismaquant
mtp
speculative-decoding
conversational
8-bit precision
License:
apache-2.0
Model card
Files
Files and versions
xet
Community
4
main
Qwen3.6-35B-A3B-PrismaQuant-4.75bit-vllm
22.9 GB
Ctrl+K
Ctrl+K
1 contributor
History:
8 commits
rdtand
README: document scale_sweep polish (closed-form AutoRound analog) + bake-off table
e347d86
verified
24 days ago
.gitattributes
Safe
1.64 kB
Upload folder using huggingface_hub
26 days ago
README.md
10.9 kB
README: document scale_sweep polish (closed-form AutoRound analog) + bake-off table
24 days ago
chat_template.jinja
Safe
7.76 kB
Upload folder using huggingface_hub
26 days ago
config.json
Safe
51.2 kB
PrismaQuant final: visual NVFP4 (108/110 DP-placed), lm_head BF16 (vLLM runtime limit), pos_embed excluded
24 days ago
configuration.json
Safe
58 Bytes
Upload folder using huggingface_hub
26 days ago
generation_config.json
Safe
202 Bytes
Upload folder using huggingface_hub
26 days ago
merges.txt
Safe
3.35 MB
Upload folder using huggingface_hub
26 days ago
mixed_native_manifest.json
Safe
487 Bytes
PrismaQuant final: visual NVFP4 (108/110 DP-placed), lm_head BF16 (vLLM runtime limit), pos_embed excluded
24 days ago
model-00001-of-00006.safetensors
4.57 GB
xet
PrismaQuant: switch polish from act_round (no-op that undid GPTQ) to closed-form scale_sweep; GPTQ + scale_sweep measured geomean out_mse 0.33 vs prior 0.99
24 days ago
model-00002-of-00006.safetensors
4.58 GB
xet
PrismaQuant: switch polish from act_round (no-op that undid GPTQ) to closed-form scale_sweep; GPTQ + scale_sweep measured geomean out_mse 0.33 vs prior 0.99
24 days ago
model-00003-of-00006.safetensors
4.58 GB
xet
PrismaQuant: switch polish from act_round (no-op that undid GPTQ) to closed-form scale_sweep; GPTQ + scale_sweep measured geomean out_mse 0.33 vs prior 0.99
24 days ago
model-00004-of-00006.safetensors
4.58 GB
xet
PrismaQuant: switch polish from act_round (no-op that undid GPTQ) to closed-form scale_sweep; GPTQ + scale_sweep measured geomean out_mse 0.33 vs prior 0.99
24 days ago
model-00005-of-00006.safetensors
4.58 GB
xet
PrismaQuant: switch polish from act_round (no-op that undid GPTQ) to closed-form scale_sweep; GPTQ + scale_sweep measured geomean out_mse 0.33 vs prior 0.99
24 days ago
model-00006-of-00006.safetensors
Safe
2.11 MB
xet
Upload folder using huggingface_hub
26 days ago
model.safetensors.index.json
Safe
14.6 MB
xet
PrismaQuant final: visual NVFP4 (108/110 DP-placed), lm_head BF16 (vLLM runtime limit), pos_embed excluded
24 days ago
preprocessor_config.json
Safe
390 Bytes
Upload folder using huggingface_hub
26 days ago
tokenizer.json
Safe
12.8 MB
xet
Upload folder using huggingface_hub
26 days ago
tokenizer_config.json
Safe
16.7 kB
Upload folder using huggingface_hub
26 days ago
video_preprocessor_config.json
Safe
385 Bytes
Upload folder using huggingface_hub
26 days ago
vocab.json
Safe
6.72 MB
Upload folder using huggingface_hub
26 days ago
