Instructions to use AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP with Transformers:

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

pipe = pipeline("image-text-to-text", model="AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP")
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("AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP")
model = AutoModelForMultimodalLM.from_pretrained("AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP", 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

llama.cpp

How to use AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M

Use Docker

docker model run hf.co/AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M

LM Studio
Jan

vLLM

How to use AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP",
		"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/AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M

SGLang

How to use AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP 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 "AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP" \
    --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": "AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP",
		"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 "AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP" \
        --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": "AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP",
		"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"
						}
					}
				]
			}
		]
	}'

Ollama
How to use AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP with Ollama:
```
ollama run hf.co/AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M
```
Unsloth Desktop

How to use AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP with Docker Model Runner:
```
docker model run hf.co/AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M
```

Lemonade

How to use AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M

Run and chat with the model

lemonade run user.Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "AMAImedia/Qwen3.6-35B-A3B-occamy-1.0-BF16-GGUF-MTP:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

AMAImedia commited on 7 days ago

Commit

1e8255a

verified ·

1 Parent(s): f4a09c7

Copy files from models/Accio-Lab/occamy-1.0-MTP

Browse files

Files changed (1) hide show

MTP/validation/native3-h200.json +480 -0

MTP/validation/native3-h200.json ADDED Viewed

	@@ -0,0 +1,480 @@

+{
+  "base_revision": "85fd90072d1a9f2432842b03588f63b745e524e4",
+  "head_sha256": "56baea0ef88de178c6c858d81adee9e532983798dce81202f21a7483e59001df",
+  "hardware": "H200 SM90",
+  "fresh_validation": {
+    "pair0": [
+      {
+        "repeat": 0,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "repeat": 1,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "cached_tokens": [
+          384,
+          384
+        ]
+      }
+    ],
+    "pair1": [
+      {
+        "repeat": 0,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "repeat": 1,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "cached_tokens": [
+          384,
+          384
+        ]
+      }
+    ]
+  },
+  "fixed_regression": {
+    "pairs": 20,
+    "tokens_exact": 20,
+    "logprobs_exact": 20,
+    "max_logprob_delta": 0.0,
+    "rows": [
+      {
+        "case": "extract",
+        "repeat": 0,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          27,
+          27
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "arithmetic",
+        "repeat": 0,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          8,
+          8
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "sort",
+        "repeat": 0,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          32,
+          32
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "tool-json",
+        "repeat": 0,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          24,
+          24
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "python-sum",
+        "repeat": 0,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          29,
+          29
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "python-fib",
+        "repeat": 0,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          65,
+          65
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "chinese",
+        "repeat": 0,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          70,
+          70
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "english",
+        "repeat": 0,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          86,
+          86
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "long-context",
+        "repeat": 0,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          9,
+          9
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "unicode",
+        "repeat": 0,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          17,
+          17
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "extract",
+        "repeat": 1,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          27,
+          27
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "arithmetic",
+        "repeat": 1,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          8,
+          8
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "sort",
+        "repeat": 1,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          32,
+          32
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "tool-json",
+        "repeat": 1,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          24,
+          24
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "python-sum",
+        "repeat": 1,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          29,
+          29
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "python-fib",
+        "repeat": 1,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          65,
+          65
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "chinese",
+        "repeat": 1,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          70,
+          70
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "english",
+        "repeat": 1,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          86,
+          86
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      },
+      {
+        "case": "long-context",
+        "repeat": 1,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          9,
+          9
+        ],
+        "cached_tokens": [
+          384,
+          384
+        ]
+      },
+      {
+        "case": "unicode",
+        "repeat": 1,
+        "tokens_exact": true,
+        "logprobs_exact": true,
+        "max_logprob_delta": 0.0,
+        "output_tokens": [
+          17,
+          17
+        ],
+        "cached_tokens": [
+          0,
+          0
+        ]
+      }
+    ]
+  },
+  "lifecycle": {
+    "baseline": {
+      "shared_prefix_independent_request_unchanged": true,
+      "two_independent_shared_requests_exact": true,
+      "finish_in_prefill": true,
+      "invalid_request_rejected": true,
+      "abort_reason": {
+        "type": "abort",
+        "message": "Aborted",
+        "status_code": null,
+        "err_type": null
+      },
+      "abort_completed": true,
+      "after_abort_and_reject_unchanged": true,
+      "health": true
+    },
+    "native-mtp3": {
+      "shared_prefix_independent_request_unchanged": true,
+      "two_independent_shared_requests_exact": true,
+      "finish_in_prefill": true,
+      "invalid_request_rejected": true,
+      "abort_reason": {
+        "type": "abort",
+        "message": "Aborted",
+        "status_code": null,
+        "err_type": null
+      },
+      "abort_completed": true,
+      "after_abort_and_reject_unchanged": true,
+      "health": true
+    }
+  },
+  "live_cache_audits": [
+    {
+      "changed_positions": [
+        0,
+        1,
+        2
+      ],
+      "released_indices": [
+        91,
+        92,
+        93
+      ],
+      "canonical_indices": [
+        1,
+        2,
+        3
+      ],
+      "released_are_free": true,
+      "canonical_not_free": true,
+      "shared_kv_unchanged": true,
+      "live_ssm_unchanged": true,
+      "live_conv_unchanged": true,
+      "lock_positive": true,
+      "cache_protected_len": 384
+    },
+    {
+      "changed_positions": [
+        0,
+        1,
+        2
+      ],
+      "released_indices": [
+        103,
+        104,
+        105
+      ],
+      "canonical_indices": [
+        1,
+        2,
+        3
+      ],
+      "released_are_free": true,
+      "canonical_not_free": true,
+      "shared_kv_unchanged": true,
+      "live_ssm_unchanged": true,
+      "live_conv_unchanged": true,
+      "lock_positive": true,
+      "cache_protected_len": 384
+    }
+  ],
+  "configuration": {
+    "baseline": {
+      "dtype": "bfloat16",
+      "tp_size": 1,
+      "disable_cuda_graph": false,
+      "disable_overlap_schedule": false,
+      "disable_radix_cache": false,
+      "skip_server_warmup": false,
+      "enable_fp32_lm_head": false,
+      "forward_hooks": null,
+      "speculative_algorithm": null,
+      "speculative_num_steps": null,
+      "speculative_num_draft_tokens": null,
+      "mamba_scheduler_strategy": "extra_buffer"
+    },
+    "native-mtp3": {
+      "dtype": "bfloat16",
+      "tp_size": 1,
+      "disable_cuda_graph": false,
+      "disable_overlap_schedule": false,
+      "disable_radix_cache": false,
+      "skip_server_warmup": false,
+      "enable_fp32_lm_head": false,
+      "forward_hooks": null,
+      "speculative_algorithm": "EAGLE",
+      "speculative_num_steps": 3,
+      "speculative_num_draft_tokens": 4,
+      "mamba_scheduler_strategy": "extra_buffer"
+    }
+  },
+  "baseline_change": {
+    "original422case_tokens_unchanged": true,
+    "cold_baseline_logprobs_max_delta_from_previous_runtime": 0.0216653011739254,
+    "warm_baseline_logprobs_unchanged": true,
+    "interpretation": "Colddecoding now starts aftercanonicalKV publication. Botharmsuse samefixedruntime; not claiming oldbaseline logprob preservation. Original9-token answer preserved."
+  },
+  "status": "passed_with_scope",
+  "scope": "H200 TP1, BF16, native three-step MTP; non-DP/non-MLP-sync, Mamba extra_buffer and radix enabled",
+  "performance": "Not measured after the final cache-boundary fix",
+  "tests": {
+    "delivered_regression_tests_passed": 6
+  },
+  "notes": "Both arms use the corrected runtime. These results do not validate quantized MTP3 or distributed serving."
+}