Downloads · 30 days
408
22% of all-time downloads
philbert440/Qwen3.8-27B-Uncensored-Cyber-NVFP4
Qwen3.8-27B-Uncensored-Cyber-NVFP4 is a image-text-to-text model from philbert440. Use it for the image-text-to-text task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
NVFP4 quant of Qwen3.8-27B-Uncensored-Cyber (v2 recipe), compressed-tensors NVFP4A16 (E2M1 4-bit weights, FP8-E4M3 scales @ group 16), for NVIDIA V100 (SM70) under 1Cat-vLLM. Vision tower and the grafted MTP head (bf1…
Downloads · 30 days
408
22% of all-time downloads
All-time downloads
1.9K
Public
Parameters
27.8B
46.2 GB on disk
Likes
8
Public
Click a slice to open those files.
.safetensors26.4 GB · 100%
How the weights are stored.
U820.3B · 73%
From the Hugging Face model README
NVFP4 quant of Qwen3.8-27B-Uncensored-Cyber (v2 recipe), compressed-tensors NVFP4A16 (E2M1 4-bit
weights, FP8-E4M3 scales @ group 16), for NVIDIA V100 (SM70) under 1Cat-vLLM.
Vision tower and the grafted MTP head (bf16) are preserved.
α=1.15 Aggressive base + a residual-cyber peel: cyber-pointed refusal direction removed by clean
norm-preserving projection (β=1.0) on the deeper layers only (apply_from=4). Fully cyber-open, general capability intact.
| cyber-open ↑ | confab ↓ | factual ↑ | gsm8k ↑ | degen ↓ | |
|---|---|---|---|---|---|
| Cyber v2 (this line) | 100/100 | 0.867 | 1.00 | 0.80 | 0.00 |
| previous Cyber build | 93/100 | 1.00 | 0.933 | 0.825 | 0.00 |
Quant: variant-C — GPTQ NVFP4A16 targeting [Linear], GDN in_proj_qkv/in_proj_z kept fp16, 768 CoT + 256 wiki calibration, actorder=weight, MSE observer. MTP head carried in bf16.
--kv-cache-dtype fp8_e5m2, MTP speculative decoding. Serves on V100/SM70 via 1Cat prepare_nvfp4_linear (min capability 70).
Uncensored / de-refused. Use responsibly and in compliance with applicable law.