Not-For-All-Audiences

abliterated

heretic

uncensor

hermes-lineage

Model card Files Files and versions

xet

Community

Instructions to use oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

Transformers

How to use oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT")
model = AutoModelForMultimodalLM.from_pretrained("oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT

SGLang

How to use oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT with Docker Model Runner:
```
docker model run hf.co/oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT
```

Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT

File size: 2,783 Bytes

a668ea0

{
  "schema_version": "q36.laptop-matrix.v1",
  "status": "planned_not_executed",
  "quants": [
    "IQ4_XS",
    "Q3_K_M",
    "IQ2_M",
    "IQ1_M"
  ],
  "backends": [
    "llama.cpp",
    "Ollama"
  ],
  "runtime_profiles": [
    "stock",
    "optimized"
  ],
  "primary_model": "Q36-v2 (same weights formerly released as v1.3)",
  "baseline_models": [
    "Huihui original",
    "Qwen/Qwen3.8-27B"
  ],
  "primary_cells": 16,
  "potential_all_model_cells": 48,
  "baseline_quant_availability": "not_verified_do_not_substitute_different_quants_silently",
  "gpu_jobs_authorized": false,
  "downloads_started": false,
  "profiles": {
    "stock": "Backend defaults explicitly recorded, with same context, prompt, output and time budgets enforced. Not an undocumented historical server setup.",
    "optimized": "Tune CPU threads, batch sizes, GPU offload and supported KV cache/flash-attention on a disjoint tuning set; freeze per device/quant/backend before test. Never optimize on scored prompts."
  },
  "common_controls": [
    "Exact weights SHA256 and projector SHA256",
    "One immutable prompt/image manifest and renderer-specific chat-template capture",
    "Separate non-thinking controlled track from publisher-recommended thinking track",
    "Same sampling, output cap, context, per-request deadline across comparable cells",
    "One warmup excluded; three text throughput repetitions; quality one trial except separately named pass@3 tests",
    "AC power, power mode, driver, runtime versions, thermals and throttling logged",
    "Randomized balanced cell order; cooldown after sustained thermal throttling",
    "No concurrent workloads; restart server between cells; clear/reuse caches explicitly",
    "Keep runtime profile independent from model identity (stock model vs stock runtime are different axes)"
  ],
  "metrics": [
    "prompt eval tokens/s",
    "decode tokens/s",
    "end-to-end tokens/s for matched IDs and length bins",
    "first visible output seconds (client) and TTFT if native provided",
    "request p50/p95 and total wall time",
    "prompt/completion token counts and missing usage",
    "process CPU percent 100%=one logical core",
    "process RSS, private/anonymous bytes if available, system commit, page faults, swap",
    "VRAM, GPU utilization, power, temperature, thermal limits",
    "content score, protocol score, refusal review, timeout/truncation/loop flags"
  ],
  "memory_policy": "Fail fast on insufficient memory; classify not runnable/OOM, not wrong answer. No model file size presented as total RAM need.",
  "cost_policy": "Preparation/API metadata/local analysis only; no inference provider, remote jobs or paid judge calls without separate authorization."
}