Image-Text-to-Text
GGUF
qwen3_5
text-generation
magicquant
mxfp4
awq
mtp
imatrix
conversational
quark
Instructions to use magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF 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 magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S # Run inference directly in the terminal: llama cli -hf magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S # Run inference directly in the terminal: llama cli -hf magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S
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 magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S # Run inference directly in the terminal: ./llama-cli -hf magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S
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 magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S # Run inference directly in the terminal: ./build/bin/llama-cli -hf magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S
Use Docker
docker model run hf.co/magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S
- LM Studio
- Jan
- vLLM
How to use magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF", "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/magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S
- Ollama
How to use magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF with Ollama:
ollama run hf.co/magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S
- Unsloth Desktop
- Pi
How to use magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF with Pi:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S
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": "magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Docker Model Runner
How to use magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF with Docker Model Runner:
docker model run hf.co/magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S
- Lemonade
How to use magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S
Run and chat with the model
lemonade run user.Qwen3.8-27B-MXFP4-MagicQuant-GGUF-UD-Q4_K_S
List all available models
lemonade list
- Hermes Agent
How to use magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF with Hermes Agent:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S
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 magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF with OpenClaw:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S
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 "magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF:UD-Q4_K_S" \ --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"
Download magicquant-manifest/magicquant.hybrid-map.json from magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF: direct link, hf CLI and curl.
- Browser
- Download file 29.3 kB
-
https://huggingface.co/magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF/resolve/3ae9728099cc03206096d3ffeeb05e8d24b5f62c/magicquant-manifest/magicquant.hybrid-map.json
- Command line
-
hf download hf://magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF@3ae9728099cc03206096d3ffeeb05e8d24b5f62c/magicquant-manifest/magicquant.hybrid-map.json
-
curl -L -o magicquant.hybrid-map.json https://huggingface.co/magiccodingman/Qwen3.8-27B-MXFP4-MagicQuant-GGUF/resolve/3ae9728099cc03206096d3ffeeb05e8d24b5f62c/magicquant-manifest/magicquant.hybrid-map.json
29.3 kB
| [ | |
| { | |
| "ExportedFileName": "Qwen3.8-27B-Quark-MXFP4-MQ-IQ4_XS_1-Generic.gguf", | |
| "DisplayName": "MQ-IQ4_XS_1", | |
| "ProviderSource": "MagicQuant recipe clone", | |
| "BaselineFamily": "IQ4_XS", | |
| "OriginalReferenceBaseline": "IQ4_NL", | |
| "SourceRecipeRepository": "magiccodingman/Qwen3.8-27B-MagicQuant-GGUF", | |
| "SourceRecipeName": "MQ-IQ4_XS_1", | |
| "SourceRun": "Unsloth Dynamic V2", | |
| "Adaptation": { | |
| "mode": "strict-downward-only", | |
| "isPureClone": false, | |
| "bf16Intermediate": false, | |
| "rule": "Never increase a source tensor's storage precision merely to match the recipe; preserve source on ties or upward requests." | |
| }, | |
| "Imatrix": { | |
| "label": "Generic", | |
| "sha256": "123a92c3ba8cd31ed2887bd348682be5b68c977b12cc30e11c632e7ddf899eaa" | |
| }, | |
| "TensorGroups": { | |
| "attention_kv": { | |
| "MXFP4": 32 | |
| }, | |
| "attention_or_ssm_output": { | |
| "MXFP4": 64 | |
| }, | |
| "attention_q_or_qkv": { | |
| "MXFP4": 64 | |
| }, | |
| "embeddings": { | |
| "Q4_K": 1 | |
| }, | |
| "ffn_down": { | |
| "MXFP4": 64 | |
| }, | |
| "ffn_up_gate": { | |
| "MXFP4": 128 | |
| }, | |
| "lm_head": { | |
| "Q6_K": 1 | |
| }, | |
| "mtp_block_64": { | |
| "F32": 7, | |
| "Q4_K": 6, | |
| "Q5_K": 1, | |
| "Q8_0": 1 | |
| }, | |
| "norms_biases_and_state": { | |
| "F32": 305 | |
| }, | |
| "other": { | |
| "F32": 48, | |
| "MXFP4": 144 | |
| } | |
| }, | |
| "EffectiveTypeCounts": { | |
| "F32": 360, | |
| "MXFP4": 496, | |
| "Q4_K": 7, | |
| "Q5_K": 1, | |
| "Q6_K": 1, | |
| "Q8_0": 1 | |
| }, | |
| "ChangedTensorCount": 10, | |
| "PreservedTensorCount": 856, | |
| "NativeMxfp4PreservedCount": 496, | |
| "MtpBlock": { | |
| "present": true, | |
| "sourceLayerCount": 1, | |
| "fullyPreservedAtSourceTypes": false, | |
| "effectiveTypeCounts": { | |
| "F32": 7, | |
| "Q4_K": 6, | |
| "Q5_K": 1, | |
| "Q8_0": 1 | |
| } | |
| }, | |
| "ActualSizeBytes": 14981769120, | |
| "ActualSizeGB": 14.98176912, | |
| "ActualSizeGiB": 13.952859789133072, | |
| "KldVsNativeMxfp4": 0.00094, | |
| "Sha256": "215dc51d41a212d58168ed80e77d0e14ca649d627563c389786263d77d1d4a3d", | |
| "VerificationPassed": true, | |
| "PublicationStatus": "published" | |
| }, | |
| { | |
| "ExportedFileName": "Qwen3.8-27B-Quark-MXFP4-UD-Q4_K_S-Unsloth.gguf", | |
| "DisplayName": "UD-Q4_K_S", | |
| "ProviderSource": "Unsloth recipe clone", | |
| "BaselineFamily": "UD-Unsloth-UD-Q4_K_S", | |
| "OriginalReferenceBaseline": "Q4_K_S", | |
| "SourceRecipeRepository": "magiccodingman/Qwen3.8-27B-MagicQuant-GGUF", | |
| "SourceRecipeName": "UD-Unsloth-UD-Q4_K_S", | |
| "SourceRun": "Unsloth Dynamic V3", | |
| "Adaptation": { | |
| "mode": "strict-downward-only", | |
| "isPureClone": false, | |
| "bf16Intermediate": false, | |
| "rule": "Never increase a source tensor's storage precision merely to match the recipe; preserve source on ties or upward requests." | |
| }, | |
| "Imatrix": { | |
| "label": "Unsloth", | |
| "sha256": "0ee5b10bd0c2fa2127c6f4b43dbfe1efd71e383b63217af9dade1de36599f1c1" | |
| }, | |
| "TensorGroups": { | |
| "attention_kv": { | |
| "MXFP4": 32 | |
| }, | |
| "attention_or_ssm_output": { | |
| "MXFP4": 64 | |
| }, | |
| "attention_q_or_qkv": { | |
| "MXFP4": 64 | |
| }, | |
| "embeddings": { | |
| "Q3_K": 1 | |
| }, | |
| "ffn_down": { | |
| "IQ3_S": 6, | |
| "IQ3_XXS": 1, | |
| "MXFP4": 56, | |
| "Q3_K": 1 | |
| }, | |
| "ffn_up_gate": { | |
| "IQ2_S": 1, | |
| "IQ2_XS": 1, | |
| "IQ3_S": 9, | |
| "IQ3_XXS": 4, | |
| "MXFP4": 102, | |
| "Q3_K": 11 | |
| }, | |
| "lm_head": { | |
| "Q6_K": 1 | |
| }, | |
| "mtp_block_64": { | |
| "F32": 7, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| }, | |
| "norms_biases_and_state": { | |
| "F32": 305 | |
| }, | |
| "other": { | |
| "F32": 48, | |
| "MXFP4": 144 | |
| } | |
| }, | |
| "EffectiveTypeCounts": { | |
| "F32": 360, | |
| "IQ2_S": 1, | |
| "IQ2_XS": 1, | |
| "IQ3_S": 15, | |
| "IQ3_XXS": 5, | |
| "MXFP4": 462, | |
| "Q3_K": 13, | |
| "Q6_K": 7, | |
| "Q8_0": 2 | |
| }, | |
| "ChangedTensorCount": 44, | |
| "PreservedTensorCount": 822, | |
| "NativeMxfp4PreservedCount": 462, | |
| "MtpBlock": { | |
| "present": true, | |
| "sourceLayerCount": 1, | |
| "fullyPreservedAtSourceTypes": false, | |
| "effectiveTypeCounts": { | |
| "F32": 7, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| } | |
| }, | |
| "ActualSizeBytes": 14547122080, | |
| "ActualSizeGB": 14.54712208, | |
| "ActualSizeGiB": 13.548063188791275, | |
| "KldVsNativeMxfp4": 0.00345, | |
| "Sha256": "b077ade0bd8aab9a1e5386ed09d72b6a24bca62fd71ec16c049b29a740f81ff3", | |
| "VerificationPassed": true, | |
| "PublicationStatus": "published" | |
| }, | |
| { | |
| "ExportedFileName": "Qwen3.8-27B-Quark-MXFP4-UD-IQ4_XS-Unsloth.gguf", | |
| "DisplayName": "UD-IQ4_XS", | |
| "ProviderSource": "Unsloth recipe clone", | |
| "BaselineFamily": "UD-Unsloth-UD-IQ4_XS", | |
| "OriginalReferenceBaseline": "IQ4_XS", | |
| "SourceRecipeRepository": "magiccodingman/Qwen3.8-27B-MagicQuant-GGUF", | |
| "SourceRecipeName": "UD-Unsloth-UD-IQ4_XS", | |
| "SourceRun": "Unsloth Dynamic V3", | |
| "Adaptation": { | |
| "mode": "strict-downward-only", | |
| "isPureClone": false, | |
| "bf16Intermediate": false, | |
| "rule": "Never increase a source tensor's storage precision merely to match the recipe; preserve source on ties or upward requests." | |
| }, | |
| "Imatrix": { | |
| "label": "Unsloth", | |
| "sha256": "0ee5b10bd0c2fa2127c6f4b43dbfe1efd71e383b63217af9dade1de36599f1c1" | |
| }, | |
| "TensorGroups": { | |
| "attention_kv": { | |
| "MXFP4": 32 | |
| }, | |
| "attention_or_ssm_output": { | |
| "MXFP4": 63, | |
| "Q3_K": 1 | |
| }, | |
| "attention_q_or_qkv": { | |
| "IQ3_S": 8, | |
| "IQ3_XXS": 3, | |
| "MXFP4": 52, | |
| "Q2_K": 1 | |
| }, | |
| "embeddings": { | |
| "Q3_K": 1 | |
| }, | |
| "ffn_down": { | |
| "IQ2_S": 1, | |
| "IQ3_S": 9, | |
| "IQ3_XXS": 5, | |
| "MXFP4": 48, | |
| "Q3_K": 1 | |
| }, | |
| "ffn_up_gate": { | |
| "IQ2_S": 6, | |
| "IQ2_XS": 2, | |
| "IQ3_S": 28, | |
| "IQ3_XXS": 9, | |
| "MXFP4": 78, | |
| "Q3_K": 5 | |
| }, | |
| "lm_head": { | |
| "Q5_K": 1 | |
| }, | |
| "mtp_block_64": { | |
| "F32": 7, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| }, | |
| "norms_biases_and_state": { | |
| "F32": 305 | |
| }, | |
| "other": { | |
| "F32": 48, | |
| "IQ3_S": 1, | |
| "MXFP4": 142, | |
| "Q3_K": 1 | |
| } | |
| }, | |
| "EffectiveTypeCounts": { | |
| "F32": 360, | |
| "IQ2_S": 7, | |
| "IQ2_XS": 2, | |
| "IQ3_S": 46, | |
| "IQ3_XXS": 17, | |
| "MXFP4": 415, | |
| "Q2_K": 1, | |
| "Q3_K": 9, | |
| "Q5_K": 1, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| }, | |
| "ChangedTensorCount": 91, | |
| "PreservedTensorCount": 775, | |
| "NativeMxfp4PreservedCount": 415, | |
| "MtpBlock": { | |
| "present": true, | |
| "sourceLayerCount": 1, | |
| "fullyPreservedAtSourceTypes": false, | |
| "effectiveTypeCounts": { | |
| "F32": 7, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| } | |
| }, | |
| "ActualSizeBytes": 13887481760, | |
| "ActualSizeGB": 13.88748176, | |
| "ActualSizeGiB": 12.933725267648697, | |
| "KldVsNativeMxfp4": 0.009018, | |
| "Sha256": "892cd2ac8aebc6f4d7cdf4d5fb0b417507a8ba9df7f038d45b9185835122cb03", | |
| "VerificationPassed": true, | |
| "PublicationStatus": "published" | |
| }, | |
| { | |
| "ExportedFileName": "Qwen3.8-27B-Quark-MXFP4-UD-Q3_K_XL-Unsloth.gguf", | |
| "DisplayName": "UD-Q3_K_XL", | |
| "ProviderSource": "Unsloth recipe clone", | |
| "BaselineFamily": "UD-Unsloth-UD-Q3_K_XL", | |
| "OriginalReferenceBaseline": "IQ3_M", | |
| "SourceRecipeRepository": "magiccodingman/Qwen3.8-27B-MagicQuant-GGUF", | |
| "SourceRecipeName": "UD-Unsloth-UD-Q3_K_XL", | |
| "SourceRun": "Unsloth Dynamic V3", | |
| "Adaptation": { | |
| "mode": "strict-downward-only", | |
| "isPureClone": false, | |
| "bf16Intermediate": false, | |
| "rule": "Never increase a source tensor's storage precision merely to match the recipe; preserve source on ties or upward requests." | |
| }, | |
| "Imatrix": { | |
| "label": "Unsloth", | |
| "sha256": "0ee5b10bd0c2fa2127c6f4b43dbfe1efd71e383b63217af9dade1de36599f1c1" | |
| }, | |
| "TensorGroups": { | |
| "attention_kv": { | |
| "MXFP4": 32 | |
| }, | |
| "attention_or_ssm_output": { | |
| "IQ3_S": 4, | |
| "IQ3_XXS": 2, | |
| "MXFP4": 54, | |
| "Q3_K": 4 | |
| }, | |
| "attention_q_or_qkv": { | |
| "IQ3_S": 29, | |
| "IQ3_XXS": 8, | |
| "MXFP4": 24, | |
| "Q2_K": 3 | |
| }, | |
| "embeddings": { | |
| "Q3_K": 1 | |
| }, | |
| "ffn_down": { | |
| "IQ2_S": 4, | |
| "IQ2_XS": 1, | |
| "IQ3_S": 17, | |
| "IQ3_XXS": 4, | |
| "MXFP4": 35, | |
| "Q3_K": 3 | |
| }, | |
| "ffn_up_gate": { | |
| "IQ2_S": 10, | |
| "IQ2_XS": 3, | |
| "IQ2_XXS": 2, | |
| "IQ3_S": 53, | |
| "IQ3_XXS": 15, | |
| "MXFP4": 41, | |
| "Q3_K": 4 | |
| }, | |
| "lm_head": { | |
| "Q5_K": 1 | |
| }, | |
| "mtp_block_64": { | |
| "F32": 7, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| }, | |
| "norms_biases_and_state": { | |
| "F32": 305 | |
| }, | |
| "other": { | |
| "F32": 48, | |
| "IQ2_S": 1, | |
| "IQ3_S": 8, | |
| "IQ3_XXS": 5, | |
| "MXFP4": 130 | |
| } | |
| }, | |
| "EffectiveTypeCounts": { | |
| "F32": 360, | |
| "IQ2_S": 15, | |
| "IQ2_XS": 4, | |
| "IQ2_XXS": 2, | |
| "IQ3_S": 111, | |
| "IQ3_XXS": 34, | |
| "MXFP4": 316, | |
| "Q2_K": 3, | |
| "Q3_K": 12, | |
| "Q5_K": 1, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| }, | |
| "ChangedTensorCount": 190, | |
| "PreservedTensorCount": 676, | |
| "NativeMxfp4PreservedCount": 316, | |
| "MtpBlock": { | |
| "present": true, | |
| "sourceLayerCount": 1, | |
| "fullyPreservedAtSourceTypes": false, | |
| "effectiveTypeCounts": { | |
| "F32": 7, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| } | |
| }, | |
| "ActualSizeBytes": 13026502560, | |
| "ActualSizeGB": 13.02650256, | |
| "ActualSizeGiB": 12.131875902414322, | |
| "KldVsNativeMxfp4": 0.022344, | |
| "Sha256": "149e7378c85768623e649afea5c9184ed1be05532948f841a39edf4870b88e63", | |
| "VerificationPassed": true, | |
| "PublicationStatus": "published" | |
| }, | |
| { | |
| "ExportedFileName": "Qwen3.8-27B-Quark-MXFP4-UD-IQ3_S-Unsloth.gguf", | |
| "DisplayName": "UD-IQ3_S", | |
| "ProviderSource": "Unsloth recipe clone", | |
| "BaselineFamily": "UD-Unsloth-UD-IQ3_S", | |
| "OriginalReferenceBaseline": "IQ3_S", | |
| "SourceRecipeRepository": "magiccodingman/Qwen3.8-27B-MagicQuant-GGUF", | |
| "SourceRecipeName": "UD-Unsloth-UD-IQ3_S", | |
| "SourceRun": "Unsloth Dynamic V3", | |
| "Adaptation": { | |
| "mode": "strict-downward-only", | |
| "isPureClone": false, | |
| "bf16Intermediate": false, | |
| "rule": "Never increase a source tensor's storage precision merely to match the recipe; preserve source on ties or upward requests." | |
| }, | |
| "Imatrix": { | |
| "label": "Unsloth", | |
| "sha256": "0ee5b10bd0c2fa2127c6f4b43dbfe1efd71e383b63217af9dade1de36599f1c1" | |
| }, | |
| "TensorGroups": { | |
| "attention_kv": { | |
| "MXFP4": 32 | |
| }, | |
| "attention_or_ssm_output": { | |
| "IQ3_S": 12, | |
| "IQ3_XXS": 3, | |
| "MXFP4": 43, | |
| "Q3_K": 6 | |
| }, | |
| "attention_q_or_qkv": { | |
| "IQ2_XXS": 1, | |
| "IQ3_S": 32, | |
| "IQ3_XXS": 21, | |
| "MXFP4": 4, | |
| "Q2_K": 6 | |
| }, | |
| "embeddings": { | |
| "Q3_K": 1 | |
| }, | |
| "ffn_down": { | |
| "IQ2_S": 6, | |
| "IQ2_XS": 1, | |
| "IQ2_XXS": 2, | |
| "IQ3_S": 23, | |
| "IQ3_XXS": 10, | |
| "MXFP4": 17, | |
| "Q3_K": 5 | |
| }, | |
| "ffn_up_gate": { | |
| "IQ1_S": 2, | |
| "IQ2_S": 14, | |
| "IQ2_XS": 10, | |
| "IQ2_XXS": 9, | |
| "IQ3_S": 45, | |
| "IQ3_XXS": 24, | |
| "MXFP4": 21, | |
| "Q3_K": 3 | |
| }, | |
| "lm_head": { | |
| "Q5_K": 1 | |
| }, | |
| "mtp_block_64": { | |
| "F32": 7, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| }, | |
| "norms_biases_and_state": { | |
| "F32": 305 | |
| }, | |
| "other": { | |
| "F32": 48, | |
| "IQ2_S": 1, | |
| "IQ2_XS": 1, | |
| "IQ3_S": 15, | |
| "IQ3_XXS": 19, | |
| "MXFP4": 107, | |
| "Q3_K": 1 | |
| } | |
| }, | |
| "EffectiveTypeCounts": { | |
| "F32": 360, | |
| "IQ1_S": 2, | |
| "IQ2_S": 21, | |
| "IQ2_XS": 12, | |
| "IQ2_XXS": 12, | |
| "IQ3_S": 127, | |
| "IQ3_XXS": 77, | |
| "MXFP4": 224, | |
| "Q2_K": 6, | |
| "Q3_K": 16, | |
| "Q5_K": 1, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| }, | |
| "ChangedTensorCount": 282, | |
| "PreservedTensorCount": 584, | |
| "NativeMxfp4PreservedCount": 224, | |
| "MtpBlock": { | |
| "present": true, | |
| "sourceLayerCount": 1, | |
| "fullyPreservedAtSourceTypes": false, | |
| "effectiveTypeCounts": { | |
| "F32": 7, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| } | |
| }, | |
| "ActualSizeBytes": 11988330400, | |
| "ActualSizeGB": 11.9883304, | |
| "ActualSizeGiB": 11.16500273346901, | |
| "KldVsNativeMxfp4": 0.042218, | |
| "Sha256": "e165d77ffd43985ebee7763e6a9642cdc8e6168ffed5017d954f6c2fc059f719", | |
| "VerificationPassed": true, | |
| "PublicationStatus": "published" | |
| }, | |
| { | |
| "ExportedFileName": "Qwen3.8-27B-Quark-MXFP4-MQ-IQ2_M_1-Generic.gguf", | |
| "DisplayName": "MQ-IQ2_M_1", | |
| "ProviderSource": "MagicQuant recipe clone", | |
| "BaselineFamily": "IQ2_M", | |
| "OriginalReferenceBaseline": "IQ3_XXS", | |
| "SourceRecipeRepository": "magiccodingman/Qwen3.8-27B-MagicQuant-GGUF", | |
| "SourceRecipeName": "MQ-IQ2_M_1", | |
| "SourceRun": "Unsloth Dynamic V2", | |
| "Adaptation": { | |
| "mode": "strict-downward-only", | |
| "isPureClone": false, | |
| "bf16Intermediate": false, | |
| "rule": "Never increase a source tensor's storage precision merely to match the recipe; preserve source on ties or upward requests." | |
| }, | |
| "Imatrix": { | |
| "label": "Generic", | |
| "sha256": "123a92c3ba8cd31ed2887bd348682be5b68c977b12cc30e11c632e7ddf899eaa" | |
| }, | |
| "TensorGroups": { | |
| "attention_kv": { | |
| "MXFP4": 32 | |
| }, | |
| "attention_or_ssm_output": { | |
| "IQ3_S": 16, | |
| "MXFP4": 48 | |
| }, | |
| "attention_q_or_qkv": { | |
| "IQ3_S": 48, | |
| "MXFP4": 16 | |
| }, | |
| "embeddings": { | |
| "Q2_K": 1 | |
| }, | |
| "ffn_down": { | |
| "IQ3_S": 64 | |
| }, | |
| "ffn_up_gate": { | |
| "IQ3_S": 64, | |
| "IQ3_XXS": 64 | |
| }, | |
| "lm_head": { | |
| "Q5_K": 1 | |
| }, | |
| "mtp_block_64": { | |
| "F32": 7, | |
| "IQ3_S": 3, | |
| "IQ4_XS": 3, | |
| "Q4_K": 2 | |
| }, | |
| "norms_biases_and_state": { | |
| "F32": 305 | |
| }, | |
| "other": { | |
| "F32": 48, | |
| "IQ1_M": 48, | |
| "IQ3_XXS": 48, | |
| "MXFP4": 48 | |
| } | |
| }, | |
| "EffectiveTypeCounts": { | |
| "F32": 360, | |
| "IQ1_M": 48, | |
| "IQ3_S": 195, | |
| "IQ3_XXS": 112, | |
| "IQ4_XS": 3, | |
| "MXFP4": 144, | |
| "Q2_K": 1, | |
| "Q4_K": 2, | |
| "Q5_K": 1 | |
| }, | |
| "ChangedTensorCount": 362, | |
| "PreservedTensorCount": 504, | |
| "NativeMxfp4PreservedCount": 144, | |
| "MtpBlock": { | |
| "present": true, | |
| "sourceLayerCount": 1, | |
| "fullyPreservedAtSourceTypes": false, | |
| "effectiveTypeCounts": { | |
| "F32": 7, | |
| "IQ3_S": 3, | |
| "IQ4_XS": 3, | |
| "Q4_K": 2 | |
| } | |
| }, | |
| "ActualSizeBytes": 11913885600, | |
| "ActualSizeGB": 11.9138856, | |
| "ActualSizeGiB": 11.095670610666275, | |
| "KldVsNativeMxfp4": 0.058879, | |
| "Sha256": "a242cddbdf9baaad660e96d26be99245f2410c154c3a0fa0c6451e3b7109eb01", | |
| "VerificationPassed": true, | |
| "PublicationStatus": "published" | |
| }, | |
| { | |
| "ExportedFileName": "Qwen3.8-27B-Quark-MXFP4-UD-IQ3_XXS-Unsloth.gguf", | |
| "DisplayName": "UD-IQ3_XXS", | |
| "ProviderSource": "Unsloth recipe clone", | |
| "BaselineFamily": "UD-Unsloth-UD-IQ3_XXS", | |
| "OriginalReferenceBaseline": "IQ3_XXS", | |
| "SourceRecipeRepository": "magiccodingman/Qwen3.8-27B-MagicQuant-GGUF", | |
| "SourceRecipeName": "UD-Unsloth-UD-IQ3_XXS", | |
| "SourceRun": "Unsloth Dynamic V3", | |
| "Adaptation": { | |
| "mode": "strict-downward-only", | |
| "isPureClone": false, | |
| "bf16Intermediate": false, | |
| "rule": "Never increase a source tensor's storage precision merely to match the recipe; preserve source on ties or upward requests." | |
| }, | |
| "Imatrix": { | |
| "label": "Unsloth", | |
| "sha256": "0ee5b10bd0c2fa2127c6f4b43dbfe1efd71e383b63217af9dade1de36599f1c1" | |
| }, | |
| "TensorGroups": { | |
| "attention_kv": { | |
| "IQ3_S": 5, | |
| "MXFP4": 27 | |
| }, | |
| "attention_or_ssm_output": { | |
| "IQ2_S": 1, | |
| "IQ2_XXS": 1, | |
| "IQ3_S": 26, | |
| "IQ3_XXS": 10, | |
| "MXFP4": 23, | |
| "Q3_K": 3 | |
| }, | |
| "attention_q_or_qkv": { | |
| "IQ2_S": 3, | |
| "IQ2_XXS": 5, | |
| "IQ3_S": 21, | |
| "IQ3_XXS": 27, | |
| "MXFP4": 2, | |
| "Q2_K": 6 | |
| }, | |
| "embeddings": { | |
| "Q2_K": 1 | |
| }, | |
| "ffn_down": { | |
| "IQ1_M": 2, | |
| "IQ1_S": 1, | |
| "IQ2_S": 7, | |
| "IQ2_XS": 4, | |
| "IQ2_XXS": 4, | |
| "IQ3_S": 20, | |
| "IQ3_XXS": 15, | |
| "MXFP4": 9, | |
| "Q2_K": 1, | |
| "Q3_K": 1 | |
| }, | |
| "ffn_up_gate": { | |
| "IQ1_M": 1, | |
| "IQ1_S": 7, | |
| "IQ2_S": 18, | |
| "IQ2_XS": 7, | |
| "IQ2_XXS": 13, | |
| "IQ3_S": 21, | |
| "IQ3_XXS": 44, | |
| "MXFP4": 15, | |
| "Q3_K": 2 | |
| }, | |
| "lm_head": { | |
| "Q4_K": 1 | |
| }, | |
| "mtp_block_64": { | |
| "F32": 7, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| }, | |
| "norms_biases_and_state": { | |
| "F32": 305 | |
| }, | |
| "other": { | |
| "F32": 48, | |
| "IQ2_S": 6, | |
| "IQ2_XS": 1, | |
| "IQ2_XXS": 1, | |
| "IQ3_S": 13, | |
| "IQ3_XXS": 24, | |
| "MXFP4": 97, | |
| "Q2_K": 2 | |
| } | |
| }, | |
| "EffectiveTypeCounts": { | |
| "F32": 360, | |
| "IQ1_M": 3, | |
| "IQ1_S": 8, | |
| "IQ2_S": 35, | |
| "IQ2_XS": 12, | |
| "IQ2_XXS": 24, | |
| "IQ3_S": 106, | |
| "IQ3_XXS": 120, | |
| "MXFP4": 173, | |
| "Q2_K": 10, | |
| "Q3_K": 6, | |
| "Q4_K": 1, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| }, | |
| "ChangedTensorCount": 333, | |
| "PreservedTensorCount": 533, | |
| "NativeMxfp4PreservedCount": 173, | |
| "MtpBlock": { | |
| "present": true, | |
| "sourceLayerCount": 1, | |
| "fullyPreservedAtSourceTypes": false, | |
| "effectiveTypeCounts": { | |
| "F32": 7, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| } | |
| }, | |
| "ActualSizeBytes": 10903975840, | |
| "ActualSizeGB": 10.90397584, | |
| "ActualSizeGiB": 10.155118852853775, | |
| "KldVsNativeMxfp4": 0.072327, | |
| "Sha256": "0cf1c2f3106833a487ccd6b1c89b185ad96dc9e9146b7d5367ab9ec797579769", | |
| "VerificationPassed": true, | |
| "PublicationStatus": "published" | |
| }, | |
| { | |
| "ExportedFileName": "Qwen3.8-27B-Quark-MXFP4-MQ-IQ2_M_2-Generic.gguf", | |
| "DisplayName": "MQ-IQ2_M_2", | |
| "ProviderSource": "MagicQuant recipe clone", | |
| "BaselineFamily": "IQ2_M", | |
| "OriginalReferenceBaseline": "IQ2_M", | |
| "SourceRecipeRepository": "magiccodingman/Qwen3.8-27B-MagicQuant-GGUF", | |
| "SourceRecipeName": "MQ-IQ2_M_2", | |
| "SourceRun": "Unsloth Dynamic V2", | |
| "Adaptation": { | |
| "mode": "strict-downward-only", | |
| "isPureClone": false, | |
| "bf16Intermediate": false, | |
| "rule": "Never increase a source tensor's storage precision merely to match the recipe; preserve source on ties or upward requests." | |
| }, | |
| "Imatrix": { | |
| "label": "Generic", | |
| "sha256": "123a92c3ba8cd31ed2887bd348682be5b68c977b12cc30e11c632e7ddf899eaa" | |
| }, | |
| "TensorGroups": { | |
| "attention_kv": { | |
| "MXFP4": 32 | |
| }, | |
| "attention_or_ssm_output": { | |
| "IQ3_S": 64 | |
| }, | |
| "attention_q_or_qkv": { | |
| "IQ3_S": 16, | |
| "IQ3_XXS": 48 | |
| }, | |
| "embeddings": { | |
| "Q2_K": 1 | |
| }, | |
| "ffn_down": { | |
| "IQ3_XXS": 64 | |
| }, | |
| "ffn_up_gate": { | |
| "IQ3_XXS": 128 | |
| }, | |
| "lm_head": { | |
| "Q3_K": 1 | |
| }, | |
| "mtp_block_64": { | |
| "F32": 7, | |
| "IQ3_S": 2, | |
| "IQ4_XS": 4, | |
| "Q5_K": 1, | |
| "Q6_K": 1 | |
| }, | |
| "norms_biases_and_state": { | |
| "F32": 305 | |
| }, | |
| "other": { | |
| "F32": 48, | |
| "IQ1_M": 48, | |
| "IQ3_XXS": 48, | |
| "MXFP4": 48 | |
| } | |
| }, | |
| "EffectiveTypeCounts": { | |
| "F32": 360, | |
| "IQ1_M": 48, | |
| "IQ3_S": 82, | |
| "IQ3_XXS": 288, | |
| "IQ4_XS": 4, | |
| "MXFP4": 80, | |
| "Q2_K": 1, | |
| "Q3_K": 1, | |
| "Q5_K": 1, | |
| "Q6_K": 1 | |
| }, | |
| "ChangedTensorCount": 426, | |
| "PreservedTensorCount": 440, | |
| "NativeMxfp4PreservedCount": 80, | |
| "MtpBlock": { | |
| "present": true, | |
| "sourceLayerCount": 1, | |
| "fullyPreservedAtSourceTypes": false, | |
| "effectiveTypeCounts": { | |
| "F32": 7, | |
| "IQ3_S": 2, | |
| "IQ4_XS": 4, | |
| "Q5_K": 1, | |
| "Q6_K": 1 | |
| } | |
| }, | |
| "ActualSizeBytes": 10691495840, | |
| "ActualSizeGB": 10.69149584, | |
| "ActualSizeGiB": 9.957231432199478, | |
| "KldVsNativeMxfp4": 0.102424, | |
| "Sha256": "a7ac992cc59c3ec95105d56c8090fd1208f0ee65bc7571d33c4202afaa8c5dc8", | |
| "VerificationPassed": true, | |
| "PublicationStatus": "published" | |
| }, | |
| { | |
| "ExportedFileName": "Qwen3.8-27B-Quark-MXFP4-UD-Q2_K_XL-Unsloth.gguf", | |
| "DisplayName": "UD-Q2_K_XL", | |
| "ProviderSource": "Unsloth recipe clone", | |
| "BaselineFamily": "UD-Unsloth-UD-Q2_K_XL", | |
| "OriginalReferenceBaseline": "IQ2_M", | |
| "SourceRecipeRepository": "magiccodingman/Qwen3.8-27B-MagicQuant-GGUF", | |
| "SourceRecipeName": "UD-Unsloth-UD-Q2_K_XL", | |
| "SourceRun": "Unsloth Dynamic V3", | |
| "Adaptation": { | |
| "mode": "strict-downward-only", | |
| "isPureClone": false, | |
| "bf16Intermediate": false, | |
| "rule": "Never increase a source tensor's storage precision merely to match the recipe; preserve source on ties or upward requests." | |
| }, | |
| "Imatrix": { | |
| "label": "Unsloth", | |
| "sha256": "0ee5b10bd0c2fa2127c6f4b43dbfe1efd71e383b63217af9dade1de36599f1c1" | |
| }, | |
| "TensorGroups": { | |
| "attention_kv": { | |
| "IQ3_S": 7, | |
| "IQ3_XXS": 1, | |
| "MXFP4": 23, | |
| "Q3_K": 1 | |
| }, | |
| "attention_or_ssm_output": { | |
| "IQ2_S": 5, | |
| "IQ2_XS": 5, | |
| "IQ2_XXS": 1, | |
| "IQ3_S": 21, | |
| "IQ3_XXS": 21, | |
| "MXFP4": 9, | |
| "Q3_K": 2 | |
| }, | |
| "attention_q_or_qkv": { | |
| "IQ1_M": 1, | |
| "IQ2_S": 12, | |
| "IQ2_XS": 6, | |
| "IQ2_XXS": 9, | |
| "IQ3_S": 4, | |
| "IQ3_XXS": 22, | |
| "Q2_K": 10 | |
| }, | |
| "embeddings": { | |
| "Q2_K": 1 | |
| }, | |
| "ffn_down": { | |
| "IQ1_S": 3, | |
| "IQ2_S": 16, | |
| "IQ2_XS": 5, | |
| "IQ2_XXS": 8, | |
| "IQ3_S": 11, | |
| "IQ3_XXS": 14, | |
| "MXFP4": 5, | |
| "Q2_K": 1, | |
| "Q3_K": 1 | |
| }, | |
| "ffn_up_gate": { | |
| "IQ1_S": 17, | |
| "IQ2_S": 26, | |
| "IQ2_XS": 14, | |
| "IQ2_XXS": 24, | |
| "IQ3_S": 12, | |
| "IQ3_XXS": 29, | |
| "MXFP4": 4, | |
| "Q2_K": 1, | |
| "Q3_K": 1 | |
| }, | |
| "lm_head": { | |
| "Q4_K": 1 | |
| }, | |
| "mtp_block_64": { | |
| "F32": 7, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| }, | |
| "norms_biases_and_state": { | |
| "F32": 305 | |
| }, | |
| "other": { | |
| "F32": 48, | |
| "IQ2_S": 8, | |
| "IQ2_XS": 4, | |
| "IQ2_XXS": 6, | |
| "IQ3_S": 2, | |
| "IQ3_XXS": 25, | |
| "MXFP4": 96, | |
| "Q2_K": 3 | |
| } | |
| }, | |
| "EffectiveTypeCounts": { | |
| "F32": 360, | |
| "IQ1_M": 1, | |
| "IQ1_S": 20, | |
| "IQ2_S": 67, | |
| "IQ2_XS": 34, | |
| "IQ2_XXS": 48, | |
| "IQ3_S": 57, | |
| "IQ3_XXS": 112, | |
| "MXFP4": 137, | |
| "Q2_K": 16, | |
| "Q3_K": 5, | |
| "Q4_K": 1, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| }, | |
| "ChangedTensorCount": 369, | |
| "PreservedTensorCount": 497, | |
| "NativeMxfp4PreservedCount": 137, | |
| "MtpBlock": { | |
| "present": true, | |
| "sourceLayerCount": 1, | |
| "fullyPreservedAtSourceTypes": false, | |
| "effectiveTypeCounts": { | |
| "F32": 7, | |
| "Q6_K": 6, | |
| "Q8_0": 2 | |
| } | |
| }, | |
| "ActualSizeBytes": 9809893280, | |
| "ActualSizeGB": 9.80989328, | |
| "ActualSizeGiB": 9.136175066232681, | |
| "KldVsNativeMxfp4": 0.111015, | |
| "Sha256": "9b899d16649813748689f2b99407b299cc9eeeae66289cf06066b050f10f3675", | |
| "VerificationPassed": true, | |
| "PublicationStatus": "published" | |
| }, | |
| { | |
| "ExportedFileName": "Qwen3.8-27B-Quark-MXFP4-UD-IQ2_XXS-Unsloth.gguf", | |
| "DisplayName": "UD-IQ2_XXS", | |
| "ProviderSource": "Unsloth recipe clone", | |
| "BaselineFamily": "UD-Unsloth-UD-IQ2_XXS", | |
| "OriginalReferenceBaseline": "IQ2_XXS", | |
| "SourceRecipeRepository": "magiccodingman/Qwen3.8-27B-MagicQuant-GGUF", | |
| "SourceRecipeName": "UD-Unsloth-UD-IQ2_XXS", | |
| "SourceRun": "Unsloth Dynamic V2", | |
| "Adaptation": { | |
| "mode": "strict-downward-only", | |
| "isPureClone": false, | |
| "bf16Intermediate": false, | |
| "rule": "Never increase a source tensor's storage precision merely to match the recipe; preserve source on ties or upward requests." | |
| }, | |
| "Imatrix": { | |
| "label": "Unsloth", | |
| "sha256": "0ee5b10bd0c2fa2127c6f4b43dbfe1efd71e383b63217af9dade1de36599f1c1" | |
| }, | |
| "TensorGroups": { | |
| "attention_kv": { | |
| "IQ3_XXS": 16, | |
| "MXFP4": 16 | |
| }, | |
| "attention_or_ssm_output": { | |
| "IQ2_S": 16, | |
| "IQ3_XXS": 48 | |
| }, | |
| "attention_q_or_qkv": { | |
| "IQ2_XS": 48, | |
| "IQ3_XXS": 16 | |
| }, | |
| "embeddings": { | |
| "Q2_K": 1 | |
| }, | |
| "ffn_down": { | |
| "IQ2_S": 64 | |
| }, | |
| "ffn_up_gate": { | |
| "IQ2_S": 128 | |
| }, | |
| "lm_head": { | |
| "Q3_K": 1 | |
| }, | |
| "mtp_block_64": { | |
| "F32": 7, | |
| "IQ4_XS": 7, | |
| "Q3_K": 1 | |
| }, | |
| "norms_biases_and_state": { | |
| "F32": 305 | |
| }, | |
| "other": { | |
| "F32": 48, | |
| "IQ1_M": 48, | |
| "IQ2_XXS": 48, | |
| "MXFP4": 48 | |
| } | |
| }, | |
| "EffectiveTypeCounts": { | |
| "F32": 360, | |
| "IQ1_M": 48, | |
| "IQ2_S": 208, | |
| "IQ2_XS": 48, | |
| "IQ2_XXS": 48, | |
| "IQ3_XXS": 80, | |
| "IQ4_XS": 7, | |
| "MXFP4": 64, | |
| "Q2_K": 1, | |
| "Q3_K": 2 | |
| }, | |
| "ChangedTensorCount": 442, | |
| "PreservedTensorCount": 424, | |
| "NativeMxfp4PreservedCount": 64, | |
| "MtpBlock": { | |
| "present": true, | |
| "sourceLayerCount": 1, | |
| "fullyPreservedAtSourceTypes": false, | |
| "effectiveTypeCounts": { | |
| "F32": 7, | |
| "IQ4_XS": 7, | |
| "Q3_K": 1 | |
| } | |
| }, | |
| "ActualSizeBytes": 9013733280, | |
| "ActualSizeGB": 9.01373328, | |
| "ActualSizeGiB": 8.394693285226822, | |
| "KldVsNativeMxfp4": 1.172122, | |
| "Sha256": "389126864bd6e3fb51639c9727567c91ebcec5e19b5c0ec072ee24d405c97f70", | |
| "VerificationPassed": true, | |
| "PublicationStatus": "removed", | |
| "RemovalReasonCode": "FAILED_QUALITY_FLOOR_AND_STRICT_DOMINANCE", | |
| "RemovalReason": "Removed from distribution after verification: KLD 1.172122 and PPL 17.87418 versus native PPL 5.801511. It is also strictly dominated by MQ-IQ2_XXS_1, which is smaller (8.22 GB) and has much lower KLD (0.321797)." | |
| }, | |
| { | |
| "ExportedFileName": "Qwen3.8-27B-Quark-MXFP4-MQ-IQ2_XXS_1-Unsloth.gguf", | |
| "DisplayName": "MQ-IQ2_XXS_1", | |
| "ProviderSource": "MagicQuant recipe clone", | |
| "BaselineFamily": "IQ2_XXS", | |
| "OriginalReferenceBaseline": "IQ2_XXS", | |
| "SourceRecipeRepository": "magiccodingman/Qwen3.8-27B-MagicQuant-GGUF", | |
| "SourceRecipeName": "MQ-IQ2_XXS_1", | |
| "SourceRun": "Unsloth Dynamic V3", | |
| "Adaptation": { | |
| "mode": "strict-downward-only", | |
| "isPureClone": false, | |
| "bf16Intermediate": false, | |
| "rule": "Never increase a source tensor's storage precision merely to match the recipe; preserve source on ties or upward requests." | |
| }, | |
| "Imatrix": { | |
| "label": "Unsloth", | |
| "sha256": "0ee5b10bd0c2fa2127c6f4b43dbfe1efd71e383b63217af9dade1de36599f1c1" | |
| }, | |
| "TensorGroups": { | |
| "attention_kv": { | |
| "IQ2_XS": 4, | |
| "IQ3_S": 6, | |
| "IQ3_XXS": 9, | |
| "MXFP4": 11, | |
| "Q2_K": 2 | |
| }, | |
| "attention_or_ssm_output": { | |
| "IQ1_M": 3, | |
| "IQ1_S": 7, | |
| "IQ2_S": 12, | |
| "IQ2_XS": 13, | |
| "IQ2_XXS": 11, | |
| "IQ3_XXS": 18 | |
| }, | |
| "attention_q_or_qkv": { | |
| "IQ1_M": 5, | |
| "IQ1_S": 5, | |
| "IQ2_S": 6, | |
| "IQ2_XS": 19, | |
| "IQ2_XXS": 20, | |
| "IQ3_S": 1, | |
| "IQ3_XXS": 5, | |
| "Q2_K": 3 | |
| }, | |
| "embeddings": { | |
| "Q2_K": 1 | |
| }, | |
| "ffn_down": { | |
| "IQ1_M": 1, | |
| "IQ1_S": 20, | |
| "IQ2_S": 7, | |
| "IQ2_XS": 2, | |
| "IQ2_XXS": 26, | |
| "IQ3_S": 3, | |
| "IQ3_XXS": 5 | |
| }, | |
| "ffn_up_gate": { | |
| "IQ1_M": 6, | |
| "IQ1_S": 53, | |
| "IQ2_S": 9, | |
| "IQ2_XS": 5, | |
| "IQ2_XXS": 45, | |
| "IQ3_XXS": 8, | |
| "MXFP4": 1, | |
| "Q3_K": 1 | |
| }, | |
| "lm_head": { | |
| "Q3_K": 1 | |
| }, | |
| "mtp_block_64": { | |
| "BF16": 8, | |
| "F32": 7 | |
| }, | |
| "norms_biases_and_state": { | |
| "F32": 305 | |
| }, | |
| "other": { | |
| "F32": 48, | |
| "IQ1_M": 4, | |
| "IQ1_S": 6, | |
| "IQ2_S": 6, | |
| "IQ2_XS": 3, | |
| "IQ2_XXS": 29, | |
| "MXFP4": 96 | |
| } | |
| }, | |
| "EffectiveTypeCounts": { | |
| "BF16": 8, | |
| "F32": 360, | |
| "IQ1_M": 19, | |
| "IQ1_S": 91, | |
| "IQ2_S": 40, | |
| "IQ2_XS": 46, | |
| "IQ2_XXS": 131, | |
| "IQ3_S": 10, | |
| "IQ3_XXS": 45, | |
| "MXFP4": 108, | |
| "Q2_K": 6, | |
| "Q3_K": 2 | |
| }, | |
| "ChangedTensorCount": 390, | |
| "PreservedTensorCount": 476, | |
| "NativeMxfp4PreservedCount": 108, | |
| "MtpBlock": { | |
| "present": true, | |
| "sourceLayerCount": 1, | |
| "fullyPreservedAtSourceTypes": true, | |
| "effectiveTypeCounts": { | |
| "BF16": 8, | |
| "F32": 7 | |
| } | |
| }, | |
| "ActualSizeBytes": 8223819680, | |
| "ActualSizeGB": 8.22381968, | |
| "ActualSizeGiB": 7.659028917551041, | |
| "KldVsNativeMxfp4": 0.321797, | |
| "Sha256": "84ab6e9cea4e1f7091e00a0b115ffa245e95f8b59e87e6e8ffdfdf3ee71dc441", | |
| "VerificationPassed": true, | |
| "PublicationStatus": "published" | |
| } | |
| ] | |