Instructions to use magiccodingman/Qwen3.8-27B-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-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-MagicQuant-GGUF:Q4_K_M # Run inference directly in the terminal: llama cli -hf magiccodingman/Qwen3.8-27B-MagicQuant-GGUF:Q4_K_M
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-MagicQuant-GGUF:Q4_K_M # Run inference directly in the terminal: llama cli -hf magiccodingman/Qwen3.8-27B-MagicQuant-GGUF: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 magiccodingman/Qwen3.8-27B-MagicQuant-GGUF:Q4_K_M # Run inference directly in the terminal: ./llama-cli -hf magiccodingman/Qwen3.8-27B-MagicQuant-GGUF: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 magiccodingman/Qwen3.8-27B-MagicQuant-GGUF:Q4_K_M # Run inference directly in the terminal: ./build/bin/llama-cli -hf magiccodingman/Qwen3.8-27B-MagicQuant-GGUF:Q4_K_M
Use Docker
docker model run hf.co/magiccodingman/Qwen3.8-27B-MagicQuant-GGUF:Q4_K_M
- LM Studio
- Jan
- vLLM
How to use magiccodingman/Qwen3.8-27B-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-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-MagicQuant-GGUF", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/magiccodingman/Qwen3.8-27B-MagicQuant-GGUF:Q4_K_M
- Ollama
How to use magiccodingman/Qwen3.8-27B-MagicQuant-GGUF with Ollama:
ollama run hf.co/magiccodingman/Qwen3.8-27B-MagicQuant-GGUF:Q4_K_M
- Unsloth Studio
How to use magiccodingman/Qwen3.8-27B-MagicQuant-GGUF with Unsloth Studio:
Install Unsloth Studio (macOS, Linux, WSL)
curl -fsSL https://unsloth.ai/install.sh | sh # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for magiccodingman/Qwen3.8-27B-MagicQuant-GGUF to start chatting
Install Unsloth Studio (Windows)
irm https://unsloth.ai/install.ps1 | iex # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for magiccodingman/Qwen3.8-27B-MagicQuant-GGUF to start chatting
Using HuggingFace Spaces for Unsloth
# No setup required # Open https://huggingface.co/spaces/unsloth/studio in your browser # Search for magiccodingman/Qwen3.8-27B-MagicQuant-GGUF to start chatting
- Pi
How to use magiccodingman/Qwen3.8-27B-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-MagicQuant-GGUF:Q4_K_M
Configure the model in Pi
# Install Pi: npm install -g @mariozechner/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-MagicQuant-GGUF:Q4_K_M" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Docker Model Runner
How to use magiccodingman/Qwen3.8-27B-MagicQuant-GGUF with Docker Model Runner:
docker model run hf.co/magiccodingman/Qwen3.8-27B-MagicQuant-GGUF:Q4_K_M
- Lemonade
How to use magiccodingman/Qwen3.8-27B-MagicQuant-GGUF with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull magiccodingman/Qwen3.8-27B-MagicQuant-GGUF:Q4_K_M
Run and chat with the model
lemonade run user.Qwen3.8-27B-MagicQuant-GGUF-Q4_K_M
List all available models
lemonade list
- Hermes Agent
How to use magiccodingman/Qwen3.8-27B-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-MagicQuant-GGUF: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 magiccodingman/Qwen3.8-27B-MagicQuant-GGUF:Q4_K_M
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use magiccodingman/Qwen3.8-27B-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-MagicQuant-GGUF: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 "magiccodingman/Qwen3.8-27B-MagicQuant-GGUF: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"
Upload folder using huggingface_hub
Browse files- .gitattributes +14 -0
- LICENSE +202 -0
- Qwen3.8-27B-LM-Q8_0.gguf +3 -0
- Qwen3.8-27B-MQ-IQ2_M_1.gguf +3 -0
- Qwen3.8-27B-MQ-IQ2_M_2.gguf +3 -0
- Qwen3.8-27B-MQ-IQ4_NL_1.gguf +3 -0
- Qwen3.8-27B-MQ-IQ4_XS_1.gguf +3 -0
- Qwen3.8-27B-MQ-Q4_K_M_1.gguf +3 -0
- Qwen3.8-27B-MQ-Q5_K_1.gguf +3 -0
- Qwen3.8-27B-MQ-Q6_K_1.gguf +3 -0
- Qwen3.8-27B-MQ-Q6_K_2.gguf +3 -0
- Qwen3.8-27B-UD-Unsloth-UD-IQ2_XXS.gguf +3 -0
- Qwen3.8-27B-UD-Unsloth-UD-Q3_K_XL.gguf +3 -0
- README.md +100 -0
- chat_template.jinja +170 -0
- config.json +140 -0
- generation_config.json +12 -0
- imatrix.dat +3 -0
- magicquant-manifest/magicquant.bad-trades.json +1176 -0
- magicquant-manifest/magicquant.clone-configs.json +0 -0
- magicquant-manifest/magicquant.final-survivors.json +495 -0
- magicquant-manifest/magicquant.hybrid-map.json +370 -0
- magicquant-manifest/magicquant.isolation-samples.json +0 -0
- magicquant-manifest/magicquant.replacements.json +1302 -0
- merges.txt +0 -0
- mmproj-BF16.gguf +3 -0
- model.safetensors.index.json +0 -0
- tokenizer.json +3 -0
- tokenizer_config.json +305 -0
- vocab.json +0 -0
.gitattributes
CHANGED
|
@@ -33,3 +33,17 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
Qwen3.8-27B-LM-Q8_0.gguf filter=lfs diff=lfs merge=lfs -text
|
| 37 |
+
Qwen3.8-27B-MQ-IQ2_M_1.gguf filter=lfs diff=lfs merge=lfs -text
|
| 38 |
+
Qwen3.8-27B-MQ-IQ2_M_2.gguf filter=lfs diff=lfs merge=lfs -text
|
| 39 |
+
Qwen3.8-27B-MQ-IQ4_NL_1.gguf filter=lfs diff=lfs merge=lfs -text
|
| 40 |
+
Qwen3.8-27B-MQ-IQ4_XS_1.gguf filter=lfs diff=lfs merge=lfs -text
|
| 41 |
+
Qwen3.8-27B-MQ-Q4_K_M_1.gguf filter=lfs diff=lfs merge=lfs -text
|
| 42 |
+
Qwen3.8-27B-MQ-Q5_K_1.gguf filter=lfs diff=lfs merge=lfs -text
|
| 43 |
+
Qwen3.8-27B-MQ-Q6_K_1.gguf filter=lfs diff=lfs merge=lfs -text
|
| 44 |
+
Qwen3.8-27B-MQ-Q6_K_2.gguf filter=lfs diff=lfs merge=lfs -text
|
| 45 |
+
Qwen3.8-27B-UD-Unsloth-UD-IQ2_XXS.gguf filter=lfs diff=lfs merge=lfs -text
|
| 46 |
+
Qwen3.8-27B-UD-Unsloth-UD-Q3_K_XL.gguf filter=lfs diff=lfs merge=lfs -text
|
| 47 |
+
imatrix.dat filter=lfs diff=lfs merge=lfs -text
|
| 48 |
+
mmproj-BF16.gguf filter=lfs diff=lfs merge=lfs -text
|
| 49 |
+
tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
LICENSE
ADDED
|
@@ -0,0 +1,202 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
Apache License
|
| 3 |
+
Version 2.0, January 2004
|
| 4 |
+
http://www.apache.org/licenses/
|
| 5 |
+
|
| 6 |
+
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
|
| 7 |
+
|
| 8 |
+
1. Definitions.
|
| 9 |
+
|
| 10 |
+
"License" shall mean the terms and conditions for use, reproduction,
|
| 11 |
+
and distribution as defined by Sections 1 through 9 of this document.
|
| 12 |
+
|
| 13 |
+
"Licensor" shall mean the copyright owner or entity authorized by
|
| 14 |
+
the copyright owner that is granting the License.
|
| 15 |
+
|
| 16 |
+
"Legal Entity" shall mean the union of the acting entity and all
|
| 17 |
+
other entities that control, are controlled by, or are under common
|
| 18 |
+
control with that entity. For the purposes of this definition,
|
| 19 |
+
"control" means (i) the power, direct or indirect, to cause the
|
| 20 |
+
direction or management of such entity, whether by contract or
|
| 21 |
+
otherwise, or (ii) ownership of fifty percent (50%) or more of the
|
| 22 |
+
outstanding shares, or (iii) beneficial ownership of such entity.
|
| 23 |
+
|
| 24 |
+
"You" (or "Your") shall mean an individual or Legal Entity
|
| 25 |
+
exercising permissions granted by this License.
|
| 26 |
+
|
| 27 |
+
"Source" form shall mean the preferred form for making modifications,
|
| 28 |
+
including but not limited to software source code, documentation
|
| 29 |
+
source, and configuration files.
|
| 30 |
+
|
| 31 |
+
"Object" form shall mean any form resulting from mechanical
|
| 32 |
+
transformation or translation of a Source form, including but
|
| 33 |
+
not limited to compiled object code, generated documentation,
|
| 34 |
+
and conversions to other media types.
|
| 35 |
+
|
| 36 |
+
"Work" shall mean the work of authorship, whether in Source or
|
| 37 |
+
Object form, made available under the License, as indicated by a
|
| 38 |
+
copyright notice that is included in or attached to the work
|
| 39 |
+
(an example is provided in the Appendix below).
|
| 40 |
+
|
| 41 |
+
"Derivative Works" shall mean any work, whether in Source or Object
|
| 42 |
+
form, that is based on (or derived from) the Work and for which the
|
| 43 |
+
editorial revisions, annotations, elaborations, or other modifications
|
| 44 |
+
represent, as a whole, an original work of authorship. For the purposes
|
| 45 |
+
of this License, Derivative Works shall not include works that remain
|
| 46 |
+
separable from, or merely link (or bind by name) to the interfaces of,
|
| 47 |
+
the Work and Derivative Works thereof.
|
| 48 |
+
|
| 49 |
+
"Contribution" shall mean any work of authorship, including
|
| 50 |
+
the original version of the Work and any modifications or additions
|
| 51 |
+
to that Work or Derivative Works thereof, that is intentionally
|
| 52 |
+
submitted to Licensor for inclusion in the Work by the copyright owner
|
| 53 |
+
or by an individual or Legal Entity authorized to submit on behalf of
|
| 54 |
+
the copyright owner. For the purposes of this definition, "submitted"
|
| 55 |
+
means any form of electronic, verbal, or written communication sent
|
| 56 |
+
to the Licensor or its representatives, including but not limited to
|
| 57 |
+
communication on electronic mailing lists, source code control systems,
|
| 58 |
+
and issue tracking systems that are managed by, or on behalf of, the
|
| 59 |
+
Licensor for the purpose of discussing and improving the Work, but
|
| 60 |
+
excluding communication that is conspicuously marked or otherwise
|
| 61 |
+
designated in writing by the copyright owner as "Not a Contribution."
|
| 62 |
+
|
| 63 |
+
"Contributor" shall mean Licensor and any individual or Legal Entity
|
| 64 |
+
on behalf of whom a Contribution has been received by Licensor and
|
| 65 |
+
subsequently incorporated within the Work.
|
| 66 |
+
|
| 67 |
+
2. Grant of Copyright License. Subject to the terms and conditions of
|
| 68 |
+
this License, each Contributor hereby grants to You a perpetual,
|
| 69 |
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
| 70 |
+
copyright license to reproduce, prepare Derivative Works of,
|
| 71 |
+
publicly display, publicly perform, sublicense, and distribute the
|
| 72 |
+
Work and such Derivative Works in Source or Object form.
|
| 73 |
+
|
| 74 |
+
3. Grant of Patent License. Subject to the terms and conditions of
|
| 75 |
+
this License, each Contributor hereby grants to You a perpetual,
|
| 76 |
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
| 77 |
+
(except as stated in this section) patent license to make, have made,
|
| 78 |
+
use, offer to sell, sell, import, and otherwise transfer the Work,
|
| 79 |
+
where such license applies only to those patent claims licensable
|
| 80 |
+
by such Contributor that are necessarily infringed by their
|
| 81 |
+
Contribution(s) alone or by combination of their Contribution(s)
|
| 82 |
+
with the Work to which such Contribution(s) was submitted. If You
|
| 83 |
+
institute patent litigation against any entity (including a
|
| 84 |
+
cross-claim or counterclaim in a lawsuit) alleging that the Work
|
| 85 |
+
or a Contribution incorporated within the Work constitutes direct
|
| 86 |
+
or contributory patent infringement, then any patent licenses
|
| 87 |
+
granted to You under this License for that Work shall terminate
|
| 88 |
+
as of the date such litigation is filed.
|
| 89 |
+
|
| 90 |
+
4. Redistribution. You may reproduce and distribute copies of the
|
| 91 |
+
Work or Derivative Works thereof in any medium, with or without
|
| 92 |
+
modifications, and in Source or Object form, provided that You
|
| 93 |
+
meet the following conditions:
|
| 94 |
+
|
| 95 |
+
(a) You must give any other recipients of the Work or
|
| 96 |
+
Derivative Works a copy of this License; and
|
| 97 |
+
|
| 98 |
+
(b) You must cause any modified files to carry prominent notices
|
| 99 |
+
stating that You changed the files; and
|
| 100 |
+
|
| 101 |
+
(c) You must retain, in the Source form of any Derivative Works
|
| 102 |
+
that You distribute, all copyright, patent, trademark, and
|
| 103 |
+
attribution notices from the Source form of the Work,
|
| 104 |
+
excluding those notices that do not pertain to any part of
|
| 105 |
+
the Derivative Works; and
|
| 106 |
+
|
| 107 |
+
(d) If the Work includes a "NOTICE" text file as part of its
|
| 108 |
+
distribution, then any Derivative Works that You distribute must
|
| 109 |
+
include a readable copy of the attribution notices contained
|
| 110 |
+
within such NOTICE file, excluding those notices that do not
|
| 111 |
+
pertain to any part of the Derivative Works, in at least one
|
| 112 |
+
of the following places: within a NOTICE text file distributed
|
| 113 |
+
as part of the Derivative Works; within the Source form or
|
| 114 |
+
documentation, if provided along with the Derivative Works; or,
|
| 115 |
+
within a display generated by the Derivative Works, if and
|
| 116 |
+
wherever such third-party notices normally appear. The contents
|
| 117 |
+
of the NOTICE file are for informational purposes only and
|
| 118 |
+
do not modify the License. You may add Your own attribution
|
| 119 |
+
notices within Derivative Works that You distribute, alongside
|
| 120 |
+
or as an addendum to the NOTICE text from the Work, provided
|
| 121 |
+
that such additional attribution notices cannot be construed
|
| 122 |
+
as modifying the License.
|
| 123 |
+
|
| 124 |
+
You may add Your own copyright statement to Your modifications and
|
| 125 |
+
may provide additional or different license terms and conditions
|
| 126 |
+
for use, reproduction, or distribution of Your modifications, or
|
| 127 |
+
for any such Derivative Works as a whole, provided Your use,
|
| 128 |
+
reproduction, and distribution of the Work otherwise complies with
|
| 129 |
+
the conditions stated in this License.
|
| 130 |
+
|
| 131 |
+
5. Submission of Contributions. Unless You explicitly state otherwise,
|
| 132 |
+
any Contribution intentionally submitted for inclusion in the Work
|
| 133 |
+
by You to the Licensor shall be under the terms and conditions of
|
| 134 |
+
this License, without any additional terms or conditions.
|
| 135 |
+
Notwithstanding the above, nothing herein shall supersede or modify
|
| 136 |
+
the terms of any separate license agreement you may have executed
|
| 137 |
+
with Licensor regarding such Contributions.
|
| 138 |
+
|
| 139 |
+
6. Trademarks. This License does not grant permission to use the trade
|
| 140 |
+
names, trademarks, service marks, or product names of the Licensor,
|
| 141 |
+
except as required for reasonable and customary use in describing the
|
| 142 |
+
origin of the Work and reproducing the content of the NOTICE file.
|
| 143 |
+
|
| 144 |
+
7. Disclaimer of Warranty. Unless required by applicable law or
|
| 145 |
+
agreed to in writing, Licensor provides the Work (and each
|
| 146 |
+
Contributor provides its Contributions) on an "AS IS" BASIS,
|
| 147 |
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
|
| 148 |
+
implied, including, without limitation, any warranties or conditions
|
| 149 |
+
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
|
| 150 |
+
PARTICULAR PURPOSE. You are solely responsible for determining the
|
| 151 |
+
appropriateness of using or redistributing the Work and assume any
|
| 152 |
+
risks associated with Your exercise of permissions under this License.
|
| 153 |
+
|
| 154 |
+
8. Limitation of Liability. In no event and under no legal theory,
|
| 155 |
+
whether in tort (including negligence), contract, or otherwise,
|
| 156 |
+
unless required by applicable law (such as deliberate and grossly
|
| 157 |
+
negligent acts) or agreed to in writing, shall any Contributor be
|
| 158 |
+
liable to You for damages, including any direct, indirect, special,
|
| 159 |
+
incidental, or consequential damages of any character arising as a
|
| 160 |
+
result of this License or out of the use or inability to use the
|
| 161 |
+
Work (including but not limited to damages for loss of goodwill,
|
| 162 |
+
work stoppage, computer failure or malfunction, or any and all
|
| 163 |
+
other commercial damages or losses), even if such Contributor
|
| 164 |
+
has been advised of the possibility of such damages.
|
| 165 |
+
|
| 166 |
+
9. Accepting Warranty or Additional Liability. While redistributing
|
| 167 |
+
the Work or Derivative Works thereof, You may choose to offer,
|
| 168 |
+
and charge a fee for, acceptance of support, warranty, indemnity,
|
| 169 |
+
or other liability obligations and/or rights consistent with this
|
| 170 |
+
License. However, in accepting such obligations, You may act only
|
| 171 |
+
on Your own behalf and on Your sole responsibility, not on behalf
|
| 172 |
+
of any other Contributor, and only if You agree to indemnify,
|
| 173 |
+
defend, and hold each Contributor harmless for any liability
|
| 174 |
+
incurred by, or claims asserted against, such Contributor by reason
|
| 175 |
+
of your accepting any such warranty or additional liability.
|
| 176 |
+
|
| 177 |
+
END OF TERMS AND CONDITIONS
|
| 178 |
+
|
| 179 |
+
APPENDIX: How to apply the Apache License to your work.
|
| 180 |
+
|
| 181 |
+
To apply the Apache License to your work, attach the following
|
| 182 |
+
boilerplate notice, with the fields enclosed by brackets "[]"
|
| 183 |
+
replaced with your own identifying information. (Don't include
|
| 184 |
+
the brackets!) The text should be enclosed in the appropriate
|
| 185 |
+
comment syntax for the file format. We also recommend that a
|
| 186 |
+
file or class name and description of purpose be included on the
|
| 187 |
+
same "printed page" as the copyright notice for easier
|
| 188 |
+
identification within third-party archives.
|
| 189 |
+
|
| 190 |
+
Copyright 2026 Alibaba Cloud
|
| 191 |
+
|
| 192 |
+
Licensed under the Apache License, Version 2.0 (the "License");
|
| 193 |
+
you may not use this file except in compliance with the License.
|
| 194 |
+
You may obtain a copy of the License at
|
| 195 |
+
|
| 196 |
+
http://www.apache.org/licenses/LICENSE-2.0
|
| 197 |
+
|
| 198 |
+
Unless required by applicable law or agreed to in writing, software
|
| 199 |
+
distributed under the License is distributed on an "AS IS" BASIS,
|
| 200 |
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 201 |
+
See the License for the specific language governing permissions and
|
| 202 |
+
limitations under the License.
|
Qwen3.8-27B-LM-Q8_0.gguf
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:09392d09c685b0f631d400b4d7e43831a70578fe6d9c250c893aabea8d2dcdf5
|
| 3 |
+
size 29047084736
|
Qwen3.8-27B-MQ-IQ2_M_1.gguf
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a66dc469f5b63509e493ce09496abc9640ed44ab5817f37ac95a6dd902aef392
|
| 3 |
+
size 11960702656
|
Qwen3.8-27B-MQ-IQ2_M_2.gguf
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cb5599696d755ad5f99a5b2b0f209ea64f618349dc4938652c298544a3567c1d
|
| 3 |
+
size 10769770176
|
Qwen3.8-27B-MQ-IQ4_NL_1.gguf
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9b2e53c95b4da92548cf06bd20ada53af634e8594a41466679ffaa668faba7c7
|
| 3 |
+
size 17062823616
|
Qwen3.8-27B-MQ-IQ4_XS_1.gguf
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c52f2eadab6708e97db3084b3ffadeeda67a845737d713fd286947dbf073a0d5
|
| 3 |
+
size 16337626816
|
Qwen3.8-27B-MQ-Q4_K_M_1.gguf
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a35b2050c4c151fab65ab42c38f3e89586cdfcc42c54b2bbe65336d7e93e15d0
|
| 3 |
+
size 18105992896
|
Qwen3.8-27B-MQ-Q5_K_1.gguf
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c7220305e3d80cc0d42e4b2b89a975786962e30378803c75e0b58021de44cd3c
|
| 3 |
+
size 20386768576
|
Qwen3.8-27B-MQ-Q6_K_1.gguf
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:49942df4f7679acfb51176bd44001730eb7e5c28c148c383649667104a1880dd
|
| 3 |
+
size 25941640896
|
Qwen3.8-27B-MQ-Q6_K_2.gguf
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ec93fca2fd079e62bcf1bd99f74afc8ebaab2736eb3481c35cacd4f960e6c980
|
| 3 |
+
size 23213909696
|
Qwen3.8-27B-UD-Unsloth-UD-IQ2_XXS.gguf
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3f189c273233b056513406a9bd75ad6c98974f3c054301ea60fdfe315b512020
|
| 3 |
+
size 9054652096
|
Qwen3.8-27B-UD-Unsloth-UD-Q3_K_XL.gguf
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3045a2778bc63fb1c58a5ee56b798fb560541b659304bdab3fa2740b2232e7e7
|
| 3 |
+
size 13481977536
|
README.md
ADDED
|
@@ -0,0 +1,100 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
tags:
|
| 4 |
+
- gguf
|
| 5 |
+
- text-generation
|
| 6 |
+
- magicquant
|
| 7 |
+
- conversational
|
| 8 |
+
base_model:
|
| 9 |
+
- Qwen/Qwen3.8-27B
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# MagicQuant Hybrids (v2.0) - Qwen3.8-27B
|
| 13 |
+
|
| 14 |
+
MagicQuant is a benchmark driven GGUF hybrid discovery and validation system focused on finding real, practical GGUF quants specific to each architecture.
|
| 15 |
+
|
| 16 |
+
Whether it's a pure baseline model built by llama.cpp, learned tensor configurations from Unsloth, or a custom built MagicQuant hybrid, the model table below shows quants that have won dominance checks, survived collapse spaces, and/or were found to be nonlinearly better. Instead of dumping every quant type possible, MagicQuant tests, validates, and brutally murders anything deemed unworthy.
|
| 17 |
+
|
| 18 |
+
<details>
|
| 19 |
+
<summary>Support MagicQuant</summary>
|
| 20 |
+
|
| 21 |
+
I’m a solo developer working full time for myself to achieve my dream. I build open source code on the side. If you like any of my work, buying me a coffee is always appreciated. Otherwise, I hope you enjoy, maybe give me a star or something. Or just send me good vibes. Either way, thank you!
|
| 22 |
+
|
| 23 |
+
[Click here to see ways to support](https://sayou.biz/support) - BTC, Paypal, GitHub sponsors.
|
| 24 |
+
|
| 25 |
+
</details>
|
| 26 |
+
|
| 27 |
+
---
|
| 28 |
+
|
| 29 |
+
## Final survivors
|
| 30 |
+
|
| 31 |
+
| Name | Provider | KLD | Size (GB) | Download |
|
| 32 |
+
|---|---|---:|---:|---|
|
| 33 |
+
| LM-Q8_0 | llama.cpp | 0.000712 | 29.05 | [Link](./../../resolve/main/Qwen3.8-27B-LM-Q8_0.gguf?download=true) |
|
| 34 |
+
| [<u>MQ-Q6_K_1</u>](#winner-notes "Replaced: Unsloth-UD-Q6_K_XL") | MagicQuant | 0.001052 | 25.94 | [Link](./../../resolve/main/Qwen3.8-27B-MQ-Q6_K_1.gguf?download=true) |
|
| 35 |
+
| [<u>MQ-Q6_K_2</u>](#winner-notes "Replaced: Unsloth-Q6_K, MQ-Unsloth-Q6_K") | MagicQuant | 0.001518 | 23.21 | [Link](./../../resolve/main/Qwen3.8-27B-MQ-Q6_K_2.gguf?download=true) |
|
| 36 |
+
| [<u>MQ-Q5_K_1</u>](#winner-notes "Replaced: Unsloth-UD-Q5_K_XL, MQ-Unsloth-Q5_K_M, Unsloth-Q5_K_M, MQ-Unsloth-Q5_K_S, Unsloth-Q5_K_S") | MagicQuant | 0.003955 | 20.39 | [Link](./../../resolve/main/Qwen3.8-27B-MQ-Q5_K_1.gguf?download=true) |
|
| 37 |
+
| [<u>MQ-Q4_K_M_1</u>](#winner-notes "Replaced: Unsloth-UD-Q4_K_XL") | MagicQuant | 0.008508 | 18.11 | [Link](./../../resolve/main/Qwen3.8-27B-MQ-Q4_K_M_1.gguf?download=true) |
|
| 38 |
+
| [<u>MQ-IQ4_NL_1</u>](#winner-notes "Replaced: Unsloth-Q4_K_M") | MagicQuant | 0.012627 | 17.06 | [Link](./../../resolve/main/Qwen3.8-27B-MQ-IQ4_NL_1.gguf?download=true) |
|
| 39 |
+
| [<u>MQ-IQ4_XS_1</u>](#winner-notes "Replaced: Unsloth-IQ4_NL, MQ-Unsloth-IQ4_XS, Unsloth-IQ4_XS") | MagicQuant | 0.013794 | 16.34 | [Link](./../../resolve/main/Qwen3.8-27B-MQ-IQ4_XS_1.gguf?download=true) |
|
| 40 |
+
| UD-Unsloth-UD-Q3_K_XL | Unsloth | 0.034902 | 13.48 | [Link](./../../resolve/main/Qwen3.8-27B-UD-Unsloth-UD-Q3_K_XL.gguf?download=true) |
|
| 41 |
+
| [<u>MQ-IQ2_M_1</u>](#winner-notes "Replaced: Unsloth-UD-IQ3_XXS") | MagicQuant | 0.057811 | 11.96 | [Link](./../../resolve/main/Qwen3.8-27B-MQ-IQ2_M_1.gguf?download=true) |
|
| 42 |
+
| [<u>MQ-IQ2_M_2</u>](#winner-notes "Replaced: Unsloth-UD-Q2_K_XL, MQ-Unsloth-UD-IQ2_M, Unsloth-UD-IQ2_M") | MagicQuant | 0.092394 | 10.77 | [Link](./../../resolve/main/Qwen3.8-27B-MQ-IQ2_M_2.gguf?download=true) |
|
| 43 |
+
| UD-Unsloth-UD-IQ2_XXS | Unsloth | 0.182364 | 9.05 | [Link](./../../resolve/main/Qwen3.8-27B-UD-Unsloth-UD-IQ2_XXS.gguf?download=true) |
|
| 44 |
+
|
| 45 |
+
<details>
|
| 46 |
+
<summary>Provider credits</summary>
|
| 47 |
+
|
| 48 |
+
- [llama.cpp](https://github.com/ggml-org/llama.cpp) — Baseline quantization formats and llama.cpp tooling.
|
| 49 |
+
- [Unsloth](https://huggingface.co/unsloth/Qwen3.8-27B-GGUF) — External learned baseline source (UD).
|
| 50 |
+
|
| 51 |
+
</details>
|
| 52 |
+
|
| 53 |
+
<details>
|
| 54 |
+
<summary>Warning - Is MagicQuant Better? (hint: how you frame the question matters)</summary>
|
| 55 |
+
|
| 56 |
+
External/custom baselines are normalized into MagicQuant's controlled comparison flow. MagicQuant rebuilds a learned baseline under native-source / MagicQuant-controlled conditions, including its own imatrix handling, so hybrids or external baselines (like Unsloth) can be judged on a more equal footing. That does **not** mean MagicQuant proved the original upstream artifact or upstream imatrix was worse. These comparisons exist for internal hybrid-search consistency and equal playing field comparisons, not as a universal judgment of the original creator's exact release artifact.
|
| 57 |
+
|
| 58 |
+
**Easier to digest explanation:**
|
| 59 |
+
|
| 60 |
+
MagicQuant compares and benchmarks the models quant to tensor configurations, but not the original artifact. And there's different reasons MagicQuant chooses to lift up a winning quant, not all winners are purely "better". It depends heavily on a variety of factors. Though choices are always documented in the repo under the manifest folder. You can always view what and why decisions were made by the automated system.
|
| 61 |
+
|
| 62 |
+
So, MagicQuant can confidently tell you, "under the same quantization to tensor configurations and identical imatrix, with this benchmark, I deemed this a winner".
|
| 63 |
+
|
| 64 |
+
</details>
|
| 65 |
+
|
| 66 |
+
<details>
|
| 67 |
+
<summary>Re-Uploading External Provider Baselines</summary>
|
| 68 |
+
|
| 69 |
+
By default, if an external provider like Unsloth is deemed the winner, the repo should generally link directly to the original provider instead of re-hosting the quant. External GGUFs are normally only re-uploaded when a specific winning variant does not already exist (e.g. Heretic models or similar).
|
| 70 |
+
|
| 71 |
+
</details>
|
| 72 |
+
|
| 73 |
+
---
|
| 74 |
+
|
| 75 |
+
## Release metadata
|
| 76 |
+
|
| 77 |
+
- [Final survivor metrics](./../../resolve/main/magicquant-manifest/magicquant.final-survivors.json) — full file names, KLD, PPL, PPL delta %, byte sizes, download targets, and replacement lineage. PPL delta % is measured against the native/reference PPL when available; negative is better and larger positive values are worse.
|
| 78 |
+
- [Hybrid tensor map](./../../resolve/main/magicquant-manifest/magicquant.hybrid-map.json) — tensor-group assignments and effective-state details for MagicQuant hybrid GGUFs.
|
| 79 |
+
- [Clone tensor configs](./../../resolve/main/magicquant-manifest/magicquant.clone-configs.json) — exact per-GGUF tensor quantization maps for reproducing this final output list in repository clone mode.
|
| 80 |
+
- [Isolation samples](./../../resolve/main/magicquant-manifest/magicquant.isolation-samples.json) — isolated base/group probe samples with KLD, PPL, PPL delta %, and size truth.
|
| 81 |
+
- [Bad trade details](./../../resolve/main/magicquant-manifest/magicquant.bad-trades.json) — structured bad-trade pruning decisions from the isolation optimizer.
|
| 82 |
+
- [Replacement details](./../../resolve/main/magicquant-manifest/magicquant.replacements.json) — structured details for baselines or anchors removed from the final download table, including reason codes, KLD deltas, PPL delta %, and size deltas.
|
| 83 |
+
|
| 84 |
+
<details>
|
| 85 |
+
<summary>Replacement reason codes</summary>
|
| 86 |
+
|
| 87 |
+
- `STRICT_DOMINANCE` — the winner was no larger and had lower real KLD than the removed anchor.
|
| 88 |
+
- `NEAR_BASELINE_PREMIUM` — the winner used only the configured near-baseline size premium and beat the real linear KLD trade line.
|
| 89 |
+
- `INTERIOR_DISCOVERY` — the winner was selected as a useful interior point inside a size/KLD gap between anchors.
|
| 90 |
+
- `SPACING_COLLAPSE` — two candidates were too close in practical output space, so the stronger one was kept.
|
| 91 |
+
- `FINAL_DOMINANCE` — a later validated survivor dominated this artifact in final real benchmark comparison.
|
| 92 |
+
|
| 93 |
+
<a id="winner-notes"></a>
|
| 94 |
+
Underlined names in the table replaced or ultimately inherited the replacement of another artifact. Hover the name for the short replacement summary, or inspect `magicquant-manifest/magicquant.replacements.json` for exact KLD/PPL/size deltas.
|
| 95 |
+
|
| 96 |
+
</details>
|
| 97 |
+
|
| 98 |
+
|
| 99 |
+
---
|
| 100 |
+
|
chat_template.jinja
ADDED
|
@@ -0,0 +1,170 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{%- set image_count = namespace(value=0) %}
|
| 2 |
+
{%- set video_count = namespace(value=0) %}
|
| 3 |
+
{%- macro render_content(content, do_vision_count, is_system_content=false) %}
|
| 4 |
+
{%- if content is string %}
|
| 5 |
+
{{- content }}
|
| 6 |
+
{%- elif content is iterable and content is not mapping %}
|
| 7 |
+
{%- for item in content %}
|
| 8 |
+
{%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
|
| 9 |
+
{%- if is_system_content %}
|
| 10 |
+
{{- raise_exception('System message cannot contain images.') }}
|
| 11 |
+
{%- endif %}
|
| 12 |
+
{%- if do_vision_count %}
|
| 13 |
+
{%- set image_count.value = image_count.value + 1 %}
|
| 14 |
+
{%- endif %}
|
| 15 |
+
{%- if add_vision_id %}
|
| 16 |
+
{{- 'Picture ' ~ image_count.value ~ ': ' }}
|
| 17 |
+
{%- endif %}
|
| 18 |
+
{{- '<|vision_start|><|image_pad|><|vision_end|>' }}
|
| 19 |
+
{%- elif 'video' in item or item.type == 'video' %}
|
| 20 |
+
{%- if is_system_content %}
|
| 21 |
+
{{- raise_exception('System message cannot contain videos.') }}
|
| 22 |
+
{%- endif %}
|
| 23 |
+
{%- if do_vision_count %}
|
| 24 |
+
{%- set video_count.value = video_count.value + 1 %}
|
| 25 |
+
{%- endif %}
|
| 26 |
+
{%- if add_vision_id %}
|
| 27 |
+
{{- 'Video ' ~ video_count.value ~ ': ' }}
|
| 28 |
+
{%- endif %}
|
| 29 |
+
{{- '<|vision_start|><|video_pad|><|vision_end|>' }}
|
| 30 |
+
{%- elif 'text' in item %}
|
| 31 |
+
{{- item.text }}
|
| 32 |
+
{%- else %}
|
| 33 |
+
{{- raise_exception('Unexpected item type in content.') }}
|
| 34 |
+
{%- endif %}
|
| 35 |
+
{%- endfor %}
|
| 36 |
+
{%- elif content is none or content is undefined %}
|
| 37 |
+
{{- '' }}
|
| 38 |
+
{%- else %}
|
| 39 |
+
{{- raise_exception('Unexpected content type.') }}
|
| 40 |
+
{%- endif %}
|
| 41 |
+
{%- endmacro %}
|
| 42 |
+
{%- if not messages %}
|
| 43 |
+
{{- raise_exception('No messages provided.') }}
|
| 44 |
+
{%- endif %}
|
| 45 |
+
{%- set reasoning_instructions = '' %}
|
| 46 |
+
{%- if enable_thinking is undefined or enable_thinking is true %}
|
| 47 |
+
{%- set resolved_reasoning_effort = reasoning_effort|default('xhigh') %}
|
| 48 |
+
{%- if resolved_reasoning_effort not in ('xhigh', 'medium', 'low') %}
|
| 49 |
+
{{- raise_exception('Unexpected reasoning effort ' ~ reasoning_effort ~ '. Supported types are xhigh (default), medium, and low.') }}
|
| 50 |
+
{%- endif %}
|
| 51 |
+
{%- if resolved_reasoning_effort == 'xhigh' %}
|
| 52 |
+
{%- set reasoning_instructions = 'Reasoning effort is set to xhigh. Please think carefully through the task, validate key assumptions, consider plausible alternatives, and prioritize correctness, consistency, and clarity in the final answer.' %}
|
| 53 |
+
{%- elif resolved_reasoning_effort == 'low' %}
|
| 54 |
+
{%- set reasoning_instructions = 'Reasoning effort is set to low. Keep your thinking brief and focused, moving directly to the conclusion without unnecessary elaboration.' %}
|
| 55 |
+
{%- endif %}
|
| 56 |
+
{%- endif %}
|
| 57 |
+
{%- if tools and tools is iterable and tools is not mapping %}
|
| 58 |
+
{{- '<|im_start|>system\n' }}
|
| 59 |
+
{%- if reasoning_instructions %}
|
| 60 |
+
{{- reasoning_instructions + '\n\n' }}
|
| 61 |
+
{%- endif %}
|
| 62 |
+
{{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
|
| 63 |
+
{%- for tool in tools %}
|
| 64 |
+
{{- "\n" }}
|
| 65 |
+
{{- tool | tojson }}
|
| 66 |
+
{%- endfor %}
|
| 67 |
+
{{- "\n</tools>" }}
|
| 68 |
+
{{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
|
| 69 |
+
{%- if messages[0].role == 'system' %}
|
| 70 |
+
{%- set content = render_content(messages[0].content, false, true)|trim %}
|
| 71 |
+
{%- if content %}
|
| 72 |
+
{{- '\n\n' + content }}
|
| 73 |
+
{%- endif %}
|
| 74 |
+
{%- endif %}
|
| 75 |
+
{{- '<|im_end|>\n' }}
|
| 76 |
+
{%- else %}
|
| 77 |
+
{%- if messages[0].role == 'system' %}
|
| 78 |
+
{%- set content = render_content(messages[0].content, false, true)|trim %}
|
| 79 |
+
{%- if content %}
|
| 80 |
+
{{- '<|im_start|>system\n' + (reasoning_instructions + '\n\n' if reasoning_instructions else '') + content + '<|im_end|>\n' }}
|
| 81 |
+
{%- elif reasoning_instructions %}
|
| 82 |
+
{{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' }}
|
| 83 |
+
{%- endif %}
|
| 84 |
+
{%- elif reasoning_instructions %}
|
| 85 |
+
{{- '<|im_start|>system\n' + reasoning_instructions + '<|im_end|>\n' }}
|
| 86 |
+
{%- endif %}
|
| 87 |
+
{%- endif %}
|
| 88 |
+
{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
|
| 89 |
+
{%- for message in messages[::-1] %}
|
| 90 |
+
{%- set index = (messages|length - 1) - loop.index0 %}
|
| 91 |
+
{%- if ns.multi_step_tool and message.role == "user" %}
|
| 92 |
+
{%- set content = render_content(message.content, false)|trim %}
|
| 93 |
+
{%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
|
| 94 |
+
{%- set ns.multi_step_tool = false %}
|
| 95 |
+
{%- set ns.last_query_index = index %}
|
| 96 |
+
{%- endif %}
|
| 97 |
+
{%- endif %}
|
| 98 |
+
{%- endfor %}
|
| 99 |
+
{%- if ns.multi_step_tool %}
|
| 100 |
+
{{- raise_exception('No user query found in messages.') }}
|
| 101 |
+
{%- endif %}
|
| 102 |
+
{%- for message in messages %}
|
| 103 |
+
{%- set content = render_content(message.content, true)|trim %}
|
| 104 |
+
{%- if message.role == "system" %}
|
| 105 |
+
{%- if not loop.first %}
|
| 106 |
+
{{- raise_exception('System message must be at the beginning.') }}
|
| 107 |
+
{%- endif %}
|
| 108 |
+
{%- elif message.role == "user" %}
|
| 109 |
+
{{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
|
| 110 |
+
{%- elif message.role == "assistant" %}
|
| 111 |
+
{%- set reasoning_content = '' %}
|
| 112 |
+
{%- if message.reasoning_content is string %}
|
| 113 |
+
{%- set reasoning_content = message.reasoning_content %}
|
| 114 |
+
{%- endif %}
|
| 115 |
+
{%- set reasoning_content = reasoning_content|trim %}
|
| 116 |
+
{%- if preserve_thinking is undefined or preserve_thinking is true or loop.index0 > ns.last_query_index %}
|
| 117 |
+
{{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
|
| 118 |
+
{%- else %}
|
| 119 |
+
{{- '<|im_start|>' + message.role + '\n' + content }}
|
| 120 |
+
{%- endif %}
|
| 121 |
+
{%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
|
| 122 |
+
{%- for tool_call in message.tool_calls %}
|
| 123 |
+
{%- if tool_call.function is defined %}
|
| 124 |
+
{%- set tool_call = tool_call.function %}
|
| 125 |
+
{%- endif %}
|
| 126 |
+
{%- if loop.first %}
|
| 127 |
+
{%- if content|trim %}
|
| 128 |
+
{{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
|
| 129 |
+
{%- else %}
|
| 130 |
+
{{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
|
| 131 |
+
{%- endif %}
|
| 132 |
+
{%- else %}
|
| 133 |
+
{{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
|
| 134 |
+
{%- endif %}
|
| 135 |
+
{%- if tool_call.arguments is defined and tool_call.arguments != '' %}
|
| 136 |
+
{%- for args_name, args_value in tool_call.arguments|items %}
|
| 137 |
+
{{- '<parameter=' + args_name + '>\n' }}
|
| 138 |
+
{%- set args_value = args_value | string if args_value is string else args_value | tojson | safe %}
|
| 139 |
+
{{- args_value }}
|
| 140 |
+
{{- '\n</parameter>\n' }}
|
| 141 |
+
{%- endfor %}
|
| 142 |
+
{%- endif %}
|
| 143 |
+
{{- '</function>\n</tool_call>' }}
|
| 144 |
+
{%- endfor %}
|
| 145 |
+
{%- endif %}
|
| 146 |
+
{{- '<|im_end|>\n' }}
|
| 147 |
+
{%- elif message.role == "tool" %}
|
| 148 |
+
{%- if loop.previtem and loop.previtem.role != "tool" %}
|
| 149 |
+
{{- '<|im_start|>user' }}
|
| 150 |
+
{%- endif %}
|
| 151 |
+
{{- '\n<tool_response>\n' }}
|
| 152 |
+
{{- content }}
|
| 153 |
+
{{- '\n</tool_response>' }}
|
| 154 |
+
{%- if not loop.last and loop.nextitem.role != "tool" %}
|
| 155 |
+
{{- '<|im_end|>\n' }}
|
| 156 |
+
{%- elif loop.last %}
|
| 157 |
+
{{- '<|im_end|>\n' }}
|
| 158 |
+
{%- endif %}
|
| 159 |
+
{%- else %}
|
| 160 |
+
{{- raise_exception('Unexpected message role.') }}
|
| 161 |
+
{%- endif %}
|
| 162 |
+
{%- endfor %}
|
| 163 |
+
{%- if add_generation_prompt %}
|
| 164 |
+
{{- '<|im_start|>assistant\n' }}
|
| 165 |
+
{%- if enable_thinking is defined and enable_thinking is false %}
|
| 166 |
+
{{- '<think>\n\n</think>\n\n' }}
|
| 167 |
+
{%- else %}
|
| 168 |
+
{{- '<think>\n' }}
|
| 169 |
+
{%- endif %}
|
| 170 |
+
{%- endif %}
|
config.json
ADDED
|
@@ -0,0 +1,140 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"Qwen3_5ForConditionalGeneration"
|
| 4 |
+
],
|
| 5 |
+
"image_token_id": 248056,
|
| 6 |
+
"language_model_only": false,
|
| 7 |
+
"model_type": "qwen3_5",
|
| 8 |
+
"text_config": {
|
| 9 |
+
"attention_bias": false,
|
| 10 |
+
"attention_dropout": 0.0,
|
| 11 |
+
"attn_output_gate": true,
|
| 12 |
+
"bos_token_id": 248044,
|
| 13 |
+
"dtype": "bfloat16",
|
| 14 |
+
"eos_token_id": 248044,
|
| 15 |
+
"full_attention_interval": 4,
|
| 16 |
+
"head_dim": 256,
|
| 17 |
+
"hidden_act": "silu",
|
| 18 |
+
"hidden_size": 5120,
|
| 19 |
+
"initializer_range": 0.02,
|
| 20 |
+
"intermediate_size": 17408,
|
| 21 |
+
"layer_types": [
|
| 22 |
+
"linear_attention",
|
| 23 |
+
"linear_attention",
|
| 24 |
+
"linear_attention",
|
| 25 |
+
"full_attention",
|
| 26 |
+
"linear_attention",
|
| 27 |
+
"linear_attention",
|
| 28 |
+
"linear_attention",
|
| 29 |
+
"full_attention",
|
| 30 |
+
"linear_attention",
|
| 31 |
+
"linear_attention",
|
| 32 |
+
"linear_attention",
|
| 33 |
+
"full_attention",
|
| 34 |
+
"linear_attention",
|
| 35 |
+
"linear_attention",
|
| 36 |
+
"linear_attention",
|
| 37 |
+
"full_attention",
|
| 38 |
+
"linear_attention",
|
| 39 |
+
"linear_attention",
|
| 40 |
+
"linear_attention",
|
| 41 |
+
"full_attention",
|
| 42 |
+
"linear_attention",
|
| 43 |
+
"linear_attention",
|
| 44 |
+
"linear_attention",
|
| 45 |
+
"full_attention",
|
| 46 |
+
"linear_attention",
|
| 47 |
+
"linear_attention",
|
| 48 |
+
"linear_attention",
|
| 49 |
+
"full_attention",
|
| 50 |
+
"linear_attention",
|
| 51 |
+
"linear_attention",
|
| 52 |
+
"linear_attention",
|
| 53 |
+
"full_attention",
|
| 54 |
+
"linear_attention",
|
| 55 |
+
"linear_attention",
|
| 56 |
+
"linear_attention",
|
| 57 |
+
"full_attention",
|
| 58 |
+
"linear_attention",
|
| 59 |
+
"linear_attention",
|
| 60 |
+
"linear_attention",
|
| 61 |
+
"full_attention",
|
| 62 |
+
"linear_attention",
|
| 63 |
+
"linear_attention",
|
| 64 |
+
"linear_attention",
|
| 65 |
+
"full_attention",
|
| 66 |
+
"linear_attention",
|
| 67 |
+
"linear_attention",
|
| 68 |
+
"linear_attention",
|
| 69 |
+
"full_attention",
|
| 70 |
+
"linear_attention",
|
| 71 |
+
"linear_attention",
|
| 72 |
+
"linear_attention",
|
| 73 |
+
"full_attention",
|
| 74 |
+
"linear_attention",
|
| 75 |
+
"linear_attention",
|
| 76 |
+
"linear_attention",
|
| 77 |
+
"full_attention",
|
| 78 |
+
"linear_attention",
|
| 79 |
+
"linear_attention",
|
| 80 |
+
"linear_attention",
|
| 81 |
+
"full_attention",
|
| 82 |
+
"linear_attention",
|
| 83 |
+
"linear_attention",
|
| 84 |
+
"linear_attention",
|
| 85 |
+
"full_attention"
|
| 86 |
+
],
|
| 87 |
+
"linear_conv_kernel_dim": 4,
|
| 88 |
+
"linear_key_head_dim": 128,
|
| 89 |
+
"linear_num_key_heads": 16,
|
| 90 |
+
"linear_num_value_heads": 48,
|
| 91 |
+
"linear_value_head_dim": 128,
|
| 92 |
+
"mamba_ssm_dtype": "float32",
|
| 93 |
+
"max_position_embeddings": 262144,
|
| 94 |
+
"model_type": "qwen3_5_text",
|
| 95 |
+
"mtp_num_hidden_layers": 1,
|
| 96 |
+
"mtp_use_dedicated_embeddings": false,
|
| 97 |
+
"num_attention_heads": 24,
|
| 98 |
+
"num_hidden_layers": 64,
|
| 99 |
+
"num_key_value_heads": 4,
|
| 100 |
+
"output_gate_type": "swish",
|
| 101 |
+
"pad_token_id": null,
|
| 102 |
+
"partial_rotary_factor": 0.25,
|
| 103 |
+
"rms_norm_eps": 1e-06,
|
| 104 |
+
"rope_parameters": {
|
| 105 |
+
"mrope_interleaved": true,
|
| 106 |
+
"mrope_section": [
|
| 107 |
+
11,
|
| 108 |
+
11,
|
| 109 |
+
10
|
| 110 |
+
],
|
| 111 |
+
"partial_rotary_factor": 0.25,
|
| 112 |
+
"rope_theta": 10000000,
|
| 113 |
+
"rope_type": "default"
|
| 114 |
+
},
|
| 115 |
+
"tie_word_embeddings": false,
|
| 116 |
+
"use_cache": true,
|
| 117 |
+
"vocab_size": 248320
|
| 118 |
+
},
|
| 119 |
+
"tie_word_embeddings": false,
|
| 120 |
+
"transformers_version": "5.8.0.dev0",
|
| 121 |
+
"video_token_id": 248057,
|
| 122 |
+
"vision_config": {
|
| 123 |
+
"deepstack_visual_indexes": [],
|
| 124 |
+
"depth": 27,
|
| 125 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 126 |
+
"hidden_size": 1152,
|
| 127 |
+
"in_channels": 3,
|
| 128 |
+
"initializer_range": 0.02,
|
| 129 |
+
"intermediate_size": 4304,
|
| 130 |
+
"model_type": "qwen3_5",
|
| 131 |
+
"num_heads": 16,
|
| 132 |
+
"num_position_embeddings": 2304,
|
| 133 |
+
"out_hidden_size": 5120,
|
| 134 |
+
"patch_size": 16,
|
| 135 |
+
"spatial_merge_size": 2,
|
| 136 |
+
"temporal_patch_size": 2
|
| 137 |
+
},
|
| 138 |
+
"vision_end_token_id": 248054,
|
| 139 |
+
"vision_start_token_id": 248053
|
| 140 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 248044,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
248046,
|
| 6 |
+
248044
|
| 7 |
+
],
|
| 8 |
+
"pad_token_id": 248044,
|
| 9 |
+
"temperature": 1.0,
|
| 10 |
+
"top_k": 20,
|
| 11 |
+
"top_p": 0.95
|
| 12 |
+
}
|
imatrix.dat
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:123a92c3ba8cd31ed2887bd348682be5b68c977b12cc30e11c632e7ddf899eaa
|
| 3 |
+
size 13642688
|
magicquant-manifest/magicquant.bad-trades.json
ADDED
|
@@ -0,0 +1,1176 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"generatedUtc": "2026-08-15T16:34:01.6732222Z",
|
| 3 |
+
"modelId": "mq-blake3:c6248ee1b24a8339716e4f620cad188eacc68755d2a1b88406a84b6249407c1d",
|
| 4 |
+
"architectureFamily": "Qwen3.8-27B",
|
| 5 |
+
"summary": {
|
| 6 |
+
"badTradeEliminations": 53,
|
| 7 |
+
"synergySecondChanceReinstatements": 2,
|
| 8 |
+
"finalKldCleanupEliminations": 0,
|
| 9 |
+
"disabledBaselines": 0,
|
| 10 |
+
"structuredBadTradeRows": 53,
|
| 11 |
+
"structuredSynergySecondChanceRows": 2
|
| 12 |
+
},
|
| 13 |
+
"thresholds": {
|
| 14 |
+
"maxSizeDeltaPercent": 4,
|
| 15 |
+
"kldMultiplier": 2.5,
|
| 16 |
+
"pplMultiplier": 3.5,
|
| 17 |
+
"floatingPointEpsilon": 1E-08
|
| 18 |
+
},
|
| 19 |
+
"badTrades": [
|
| 20 |
+
{
|
| 21 |
+
"Scope": "group",
|
| 22 |
+
"GroupName": "embeddings",
|
| 23 |
+
"RemovedCandidate": "Unsloth-Q4_K_M",
|
| 24 |
+
"AcceptedAnchor": "Unsloth-Q5_K_M",
|
| 25 |
+
"Reason": "Reason: small size gain (0.32%) but disproportionate damage (KLD x1.46, |PPL| x6.07).",
|
| 26 |
+
"RemovedSizeBytes": 49962325696,
|
| 27 |
+
"RemovedSizeGB": 49.962325696,
|
| 28 |
+
"RemovedSizeGiB": 46.5310418009758,
|
| 29 |
+
"RemovedKld": 0.000401,
|
| 30 |
+
"RemovedPplDeltaPercent": 0.007900133749438237,
|
| 31 |
+
"AnchorSizeBytes": 50121250496,
|
| 32 |
+
"AnchorSizeGB": 50.121250496,
|
| 33 |
+
"AnchorSizeGiB": 46.67905205488205,
|
| 34 |
+
"AnchorKld": 0.000274,
|
| 35 |
+
"AnchorPplDeltaPercent": 0.0013018279090581157,
|
| 36 |
+
"SizeDeltaPercent": 0.31708067621473895,
|
| 37 |
+
"KldRatio": 1.4635036496350364,
|
| 38 |
+
"PplAbsRatio": 6.0684931506454305
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"Scope": "group",
|
| 42 |
+
"GroupName": "embeddings",
|
| 43 |
+
"RemovedCandidate": "Unsloth-UD-Q4_K_XL",
|
| 44 |
+
"AcceptedAnchor": "Unsloth-Q5_K_M",
|
| 45 |
+
"Reason": "Reason: small size gain (0.32%) but disproportionate damage (KLD x1.46, |PPL| x6.07).",
|
| 46 |
+
"RemovedSizeBytes": 49962325696,
|
| 47 |
+
"RemovedSizeGB": 49.962325696,
|
| 48 |
+
"RemovedSizeGiB": 46.5310418009758,
|
| 49 |
+
"RemovedKld": 0.000401,
|
| 50 |
+
"RemovedPplDeltaPercent": 0.007900133749438237,
|
| 51 |
+
"AnchorSizeBytes": 50121250496,
|
| 52 |
+
"AnchorSizeGB": 50.121250496,
|
| 53 |
+
"AnchorSizeGiB": 46.67905205488205,
|
| 54 |
+
"AnchorKld": 0.000274,
|
| 55 |
+
"AnchorPplDeltaPercent": 0.0013018279090581157,
|
| 56 |
+
"SizeDeltaPercent": 0.31708067621473895,
|
| 57 |
+
"KldRatio": 1.4635036496350364,
|
| 58 |
+
"PplAbsRatio": 6.0684931506454305
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"Scope": "group",
|
| 62 |
+
"GroupName": "embeddings",
|
| 63 |
+
"RemovedCandidate": "Unsloth-Q4_K_S",
|
| 64 |
+
"AcceptedAnchor": "Unsloth-Q5_K_M",
|
| 65 |
+
"Reason": "Reason: small size gain (0.32%) but disproportionate damage (KLD x1.46, |PPL| x6.07).",
|
| 66 |
+
"RemovedSizeBytes": 49962325696,
|
| 67 |
+
"RemovedSizeGB": 49.962325696,
|
| 68 |
+
"RemovedSizeGiB": 46.5310418009758,
|
| 69 |
+
"RemovedKld": 0.000401,
|
| 70 |
+
"RemovedPplDeltaPercent": 0.007900133749438237,
|
| 71 |
+
"AnchorSizeBytes": 50121250496,
|
| 72 |
+
"AnchorSizeGB": 50.121250496,
|
| 73 |
+
"AnchorSizeGiB": 46.67905205488205,
|
| 74 |
+
"AnchorKld": 0.000274,
|
| 75 |
+
"AnchorPplDeltaPercent": 0.0013018279090581157,
|
| 76 |
+
"SizeDeltaPercent": 0.31708067621473895,
|
| 77 |
+
"KldRatio": 1.4635036496350364,
|
| 78 |
+
"PplAbsRatio": 6.0684931506454305
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"Scope": "group",
|
| 82 |
+
"GroupName": "embeddings",
|
| 83 |
+
"RemovedCandidate": "Unsloth-IQ4_NL",
|
| 84 |
+
"AcceptedAnchor": "Unsloth-Q5_K_M",
|
| 85 |
+
"Reason": "Reason: small size gain (0.32%) but disproportionate damage (KLD x1.46, |PPL| x6.07).",
|
| 86 |
+
"RemovedSizeBytes": 49962325696,
|
| 87 |
+
"RemovedSizeGB": 49.962325696,
|
| 88 |
+
"RemovedSizeGiB": 46.5310418009758,
|
| 89 |
+
"RemovedKld": 0.000401,
|
| 90 |
+
"RemovedPplDeltaPercent": 0.007900133749438237,
|
| 91 |
+
"AnchorSizeBytes": 50121250496,
|
| 92 |
+
"AnchorSizeGB": 50.121250496,
|
| 93 |
+
"AnchorSizeGiB": 46.67905205488205,
|
| 94 |
+
"AnchorKld": 0.000274,
|
| 95 |
+
"AnchorPplDeltaPercent": 0.0013018279090581157,
|
| 96 |
+
"SizeDeltaPercent": 0.31708067621473895,
|
| 97 |
+
"KldRatio": 1.4635036496350364,
|
| 98 |
+
"PplAbsRatio": 6.0684931506454305
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"Scope": "group",
|
| 102 |
+
"GroupName": "embeddings",
|
| 103 |
+
"RemovedCandidate": "Unsloth-IQ4_XS",
|
| 104 |
+
"AcceptedAnchor": "Unsloth-Q5_K_M",
|
| 105 |
+
"Reason": "Reason: small size gain (0.32%) but disproportionate damage (KLD x1.46, |PPL| x6.07).",
|
| 106 |
+
"RemovedSizeBytes": 49962325696,
|
| 107 |
+
"RemovedSizeGB": 49.962325696,
|
| 108 |
+
"RemovedSizeGiB": 46.5310418009758,
|
| 109 |
+
"RemovedKld": 0.000401,
|
| 110 |
+
"RemovedPplDeltaPercent": 0.007900133749438237,
|
| 111 |
+
"AnchorSizeBytes": 50121250496,
|
| 112 |
+
"AnchorSizeGB": 50.121250496,
|
| 113 |
+
"AnchorSizeGiB": 46.67905205488205,
|
| 114 |
+
"AnchorKld": 0.000274,
|
| 115 |
+
"AnchorPplDeltaPercent": 0.0013018279090581157,
|
| 116 |
+
"SizeDeltaPercent": 0.31708067621473895,
|
| 117 |
+
"KldRatio": 1.4635036496350364,
|
| 118 |
+
"PplAbsRatio": 6.0684931506454305
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"Scope": "group",
|
| 122 |
+
"GroupName": "embeddings",
|
| 123 |
+
"RemovedCandidate": "Unsloth-Q3_K_M",
|
| 124 |
+
"AcceptedAnchor": "Unsloth-Q5_K_M",
|
| 125 |
+
"Reason": "Reason: small size gain (0.65%) but disproportionate damage (KLD x3.19, |PPL| x39.63).",
|
| 126 |
+
"RemovedSizeBytes": 49793468096,
|
| 127 |
+
"RemovedSizeGB": 49.793468096000005,
|
| 128 |
+
"RemovedSizeGiB": 46.37378090620041,
|
| 129 |
+
"RemovedKld": 0.000873,
|
| 130 |
+
"RemovedPplDeltaPercent": -0.05159161836826138,
|
| 131 |
+
"AnchorSizeBytes": 50121250496,
|
| 132 |
+
"AnchorSizeGB": 50.121250496,
|
| 133 |
+
"AnchorSizeGiB": 46.67905205488205,
|
| 134 |
+
"AnchorKld": 0.000274,
|
| 135 |
+
"AnchorPplDeltaPercent": 0.0013018279090581157,
|
| 136 |
+
"SizeDeltaPercent": 0.653978894692899,
|
| 137 |
+
"KldRatio": 3.1861313868613137,
|
| 138 |
+
"PplAbsRatio": 39.63013698606937
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"Scope": "group",
|
| 142 |
+
"GroupName": "embeddings",
|
| 143 |
+
"RemovedCandidate": "Unsloth-Q3_K_S",
|
| 144 |
+
"AcceptedAnchor": "Unsloth-Q5_K_M",
|
| 145 |
+
"Reason": "Reason: small size gain (0.65%) but disproportionate damage (KLD x3.19, |PPL| x39.63).",
|
| 146 |
+
"RemovedSizeBytes": 49793468096,
|
| 147 |
+
"RemovedSizeGB": 49.793468096000005,
|
| 148 |
+
"RemovedSizeGiB": 46.37378090620041,
|
| 149 |
+
"RemovedKld": 0.000873,
|
| 150 |
+
"RemovedPplDeltaPercent": -0.05159161836826138,
|
| 151 |
+
"AnchorSizeBytes": 50121250496,
|
| 152 |
+
"AnchorSizeGB": 50.121250496,
|
| 153 |
+
"AnchorSizeGiB": 46.67905205488205,
|
| 154 |
+
"AnchorKld": 0.000274,
|
| 155 |
+
"AnchorPplDeltaPercent": 0.0013018279090581157,
|
| 156 |
+
"SizeDeltaPercent": 0.653978894692899,
|
| 157 |
+
"KldRatio": 3.1861313868613137,
|
| 158 |
+
"PplAbsRatio": 39.63013698606937
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"Scope": "group",
|
| 162 |
+
"GroupName": "embeddings",
|
| 163 |
+
"RemovedCandidate": "Unsloth-UD-Q3_K_XL",
|
| 164 |
+
"AcceptedAnchor": "Unsloth-Q5_K_M",
|
| 165 |
+
"Reason": "Reason: small size gain (0.65%) but disproportionate damage (KLD x3.19, |PPL| x39.63).",
|
| 166 |
+
"RemovedSizeBytes": 49793468096,
|
| 167 |
+
"RemovedSizeGB": 49.793468096000005,
|
| 168 |
+
"RemovedSizeGiB": 46.37378090620041,
|
| 169 |
+
"RemovedKld": 0.000873,
|
| 170 |
+
"RemovedPplDeltaPercent": -0.05159161836826138,
|
| 171 |
+
"AnchorSizeBytes": 50121250496,
|
| 172 |
+
"AnchorSizeGB": 50.121250496,
|
| 173 |
+
"AnchorSizeGiB": 46.67905205488205,
|
| 174 |
+
"AnchorKld": 0.000274,
|
| 175 |
+
"AnchorPplDeltaPercent": 0.0013018279090581157,
|
| 176 |
+
"SizeDeltaPercent": 0.653978894692899,
|
| 177 |
+
"KldRatio": 3.1861313868613137,
|
| 178 |
+
"PplAbsRatio": 39.63013698606937
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"Scope": "group",
|
| 182 |
+
"GroupName": "embeddings",
|
| 183 |
+
"RemovedCandidate": "Unsloth-UD-IQ3_XXS",
|
| 184 |
+
"AcceptedAnchor": "Unsloth-Q5_K_M",
|
| 185 |
+
"Reason": "Reason: small size gain (0.91%) but disproportionate damage (KLD x7.11, |PPL| x1.04).",
|
| 186 |
+
"RemovedSizeBytes": 49664341696,
|
| 187 |
+
"RemovedSizeGB": 49.664341696,
|
| 188 |
+
"RemovedSizeGiB": 46.25352257490158,
|
| 189 |
+
"RemovedKld": 0.001949,
|
| 190 |
+
"RemovedPplDeltaPercent": -0.0013553276861340167,
|
| 191 |
+
"AnchorSizeBytes": 50121250496,
|
| 192 |
+
"AnchorSizeGB": 50.121250496,
|
| 193 |
+
"AnchorSizeGiB": 46.67905205488205,
|
| 194 |
+
"AnchorKld": 0.000274,
|
| 195 |
+
"AnchorPplDeltaPercent": 0.0013018279090581157,
|
| 196 |
+
"SizeDeltaPercent": 0.9116069441173744,
|
| 197 |
+
"KldRatio": 7.1131386861313874,
|
| 198 |
+
"PplAbsRatio": 1.0410958904042922
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"Scope": "group",
|
| 202 |
+
"GroupName": "embeddings",
|
| 203 |
+
"RemovedCandidate": "Unsloth-UD-IQ2_M",
|
| 204 |
+
"AcceptedAnchor": "Unsloth-Q5_K_M",
|
| 205 |
+
"Reason": "Reason: small size gain (0.91%) but disproportionate damage (KLD x7.11, |PPL| x1.04).",
|
| 206 |
+
"RemovedSizeBytes": 49664341696,
|
| 207 |
+
"RemovedSizeGB": 49.664341696,
|
| 208 |
+
"RemovedSizeGiB": 46.25352257490158,
|
| 209 |
+
"RemovedKld": 0.001949,
|
| 210 |
+
"RemovedPplDeltaPercent": -0.0013553276861340167,
|
| 211 |
+
"AnchorSizeBytes": 50121250496,
|
| 212 |
+
"AnchorSizeGB": 50.121250496,
|
| 213 |
+
"AnchorSizeGiB": 46.67905205488205,
|
| 214 |
+
"AnchorKld": 0.000274,
|
| 215 |
+
"AnchorPplDeltaPercent": 0.0013018279090581157,
|
| 216 |
+
"SizeDeltaPercent": 0.9116069441173744,
|
| 217 |
+
"KldRatio": 7.1131386861313874,
|
| 218 |
+
"PplAbsRatio": 1.0410958904042922
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"Scope": "group",
|
| 222 |
+
"GroupName": "embeddings",
|
| 223 |
+
"RemovedCandidate": "Unsloth-UD-Q2_K_XL",
|
| 224 |
+
"AcceptedAnchor": "Unsloth-Q5_K_M",
|
| 225 |
+
"Reason": "Reason: small size gain (0.91%) but disproportionate damage (KLD x7.11, |PPL| x1.04).",
|
| 226 |
+
"RemovedSizeBytes": 49664341696,
|
| 227 |
+
"RemovedSizeGB": 49.664341696,
|
| 228 |
+
"RemovedSizeGiB": 46.25352257490158,
|
| 229 |
+
"RemovedKld": 0.001949,
|
| 230 |
+
"RemovedPplDeltaPercent": -0.0013553276861340167,
|
| 231 |
+
"AnchorSizeBytes": 50121250496,
|
| 232 |
+
"AnchorSizeGB": 50.121250496,
|
| 233 |
+
"AnchorSizeGiB": 46.67905205488205,
|
| 234 |
+
"AnchorKld": 0.000274,
|
| 235 |
+
"AnchorPplDeltaPercent": 0.0013018279090581157,
|
| 236 |
+
"SizeDeltaPercent": 0.9116069441173744,
|
| 237 |
+
"KldRatio": 7.1131386861313874,
|
| 238 |
+
"PplAbsRatio": 1.0410958904042922
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"Scope": "group",
|
| 242 |
+
"GroupName": "embeddings",
|
| 243 |
+
"RemovedCandidate": "Unsloth-UD-IQ2_XXS",
|
| 244 |
+
"AcceptedAnchor": "Unsloth-Q5_K_M",
|
| 245 |
+
"Reason": "Reason: small size gain (0.91%) but disproportionate damage (KLD x7.11, |PPL| x1.04).",
|
| 246 |
+
"RemovedSizeBytes": 49664341696,
|
| 247 |
+
"RemovedSizeGB": 49.664341696,
|
| 248 |
+
"RemovedSizeGiB": 46.25352257490158,
|
| 249 |
+
"RemovedKld": 0.001949,
|
| 250 |
+
"RemovedPplDeltaPercent": -0.0013553276861340167,
|
| 251 |
+
"AnchorSizeBytes": 50121250496,
|
| 252 |
+
"AnchorSizeGB": 50.121250496,
|
| 253 |
+
"AnchorSizeGiB": 46.67905205488205,
|
| 254 |
+
"AnchorKld": 0.000274,
|
| 255 |
+
"AnchorPplDeltaPercent": 0.0013018279090581157,
|
| 256 |
+
"SizeDeltaPercent": 0.9116069441173744,
|
| 257 |
+
"KldRatio": 7.1131386861313874,
|
| 258 |
+
"PplAbsRatio": 1.0410958904042922
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"Scope": "group",
|
| 262 |
+
"GroupName": "lm_head",
|
| 263 |
+
"RemovedCandidate": "Unsloth-UD-Q3_K_XL",
|
| 264 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 265 |
+
"Reason": "Reason: small size gain (0.34%) but disproportionate damage (KLD x2.65, |PPL| x2.32).",
|
| 266 |
+
"RemovedSizeBytes": 50121250496,
|
| 267 |
+
"RemovedSizeGB": 50.121250496,
|
| 268 |
+
"RemovedSizeGiB": 46.67905205488205,
|
| 269 |
+
"RemovedKld": 0.00124,
|
| 270 |
+
"RemovedPplDeltaPercent": 0.13134195274186475,
|
| 271 |
+
"AnchorSizeBytes": 50290108096,
|
| 272 |
+
"AnchorSizeGB": 50.290108096000004,
|
| 273 |
+
"AnchorSizeGiB": 46.83631294965744,
|
| 274 |
+
"AnchorKld": 0.000468,
|
| 275 |
+
"AnchorPplDeltaPercent": -0.05670976370930926,
|
| 276 |
+
"SizeDeltaPercent": 0.33576702535151376,
|
| 277 |
+
"KldRatio": 2.6495726495726495,
|
| 278 |
+
"PplAbsRatio": 2.316037735849429
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"Scope": "group",
|
| 282 |
+
"GroupName": "lm_head",
|
| 283 |
+
"RemovedCandidate": "Unsloth-UD-IQ3_XXS",
|
| 284 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 285 |
+
"Reason": "Reason: small size gain (0.34%) but disproportionate damage (KLD x2.65, |PPL| x2.32).",
|
| 286 |
+
"RemovedSizeBytes": 50121250496,
|
| 287 |
+
"RemovedSizeGB": 50.121250496,
|
| 288 |
+
"RemovedSizeGiB": 46.67905205488205,
|
| 289 |
+
"RemovedKld": 0.00124,
|
| 290 |
+
"RemovedPplDeltaPercent": 0.13134195274186475,
|
| 291 |
+
"AnchorSizeBytes": 50290108096,
|
| 292 |
+
"AnchorSizeGB": 50.290108096000004,
|
| 293 |
+
"AnchorSizeGiB": 46.83631294965744,
|
| 294 |
+
"AnchorKld": 0.000468,
|
| 295 |
+
"AnchorPplDeltaPercent": -0.05670976370930926,
|
| 296 |
+
"SizeDeltaPercent": 0.33576702535151376,
|
| 297 |
+
"KldRatio": 2.6495726495726495,
|
| 298 |
+
"PplAbsRatio": 2.316037735849429
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"Scope": "group",
|
| 302 |
+
"GroupName": "lm_head",
|
| 303 |
+
"RemovedCandidate": "Unsloth-UD-IQ2_M",
|
| 304 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 305 |
+
"Reason": "Reason: small size gain (0.99%) but disproportionate damage (KLD x35.33, |PPL| x32.48).",
|
| 306 |
+
"RemovedSizeBytes": 49793468096,
|
| 307 |
+
"RemovedSizeGB": 49.793468096000005,
|
| 308 |
+
"RemovedSizeGiB": 46.37378090620041,
|
| 309 |
+
"RemovedKld": 0.016535,
|
| 310 |
+
"RemovedPplDeltaPercent": 1.8420864913062889,
|
| 311 |
+
"AnchorSizeBytes": 50290108096,
|
| 312 |
+
"AnchorSizeGB": 50.290108096000004,
|
| 313 |
+
"AnchorSizeGiB": 46.83631294965744,
|
| 314 |
+
"AnchorKld": 0.000468,
|
| 315 |
+
"AnchorPplDeltaPercent": -0.05670976370930926,
|
| 316 |
+
"SizeDeltaPercent": 0.9875500745632757,
|
| 317 |
+
"KldRatio": 35.331196581196586,
|
| 318 |
+
"PplAbsRatio": 32.4827044025207
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"Scope": "group",
|
| 322 |
+
"GroupName": "lm_head",
|
| 323 |
+
"RemovedCandidate": "Unsloth-UD-Q2_K_XL",
|
| 324 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 325 |
+
"Reason": "Reason: small size gain (0.99%) but disproportionate damage (KLD x35.33, |PPL| x32.48).",
|
| 326 |
+
"RemovedSizeBytes": 49793468096,
|
| 327 |
+
"RemovedSizeGB": 49.793468096000005,
|
| 328 |
+
"RemovedSizeGiB": 46.37378090620041,
|
| 329 |
+
"RemovedKld": 0.016535,
|
| 330 |
+
"RemovedPplDeltaPercent": 1.8420864913062889,
|
| 331 |
+
"AnchorSizeBytes": 50290108096,
|
| 332 |
+
"AnchorSizeGB": 50.290108096000004,
|
| 333 |
+
"AnchorSizeGiB": 46.83631294965744,
|
| 334 |
+
"AnchorKld": 0.000468,
|
| 335 |
+
"AnchorPplDeltaPercent": -0.05670976370930926,
|
| 336 |
+
"SizeDeltaPercent": 0.9875500745632757,
|
| 337 |
+
"KldRatio": 35.331196581196586,
|
| 338 |
+
"PplAbsRatio": 32.4827044025207
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"Scope": "group",
|
| 342 |
+
"GroupName": "lm_head",
|
| 343 |
+
"RemovedCandidate": "Unsloth-UD-IQ2_XXS",
|
| 344 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 345 |
+
"Reason": "Reason: small size gain (0.99%) but disproportionate damage (KLD x35.33, |PPL| x32.48).",
|
| 346 |
+
"RemovedSizeBytes": 49793468096,
|
| 347 |
+
"RemovedSizeGB": 49.793468096000005,
|
| 348 |
+
"RemovedSizeGiB": 46.37378090620041,
|
| 349 |
+
"RemovedKld": 0.016535,
|
| 350 |
+
"RemovedPplDeltaPercent": 1.8420864913062889,
|
| 351 |
+
"AnchorSizeBytes": 50290108096,
|
| 352 |
+
"AnchorSizeGB": 50.290108096000004,
|
| 353 |
+
"AnchorSizeGiB": 46.83631294965744,
|
| 354 |
+
"AnchorKld": 0.000468,
|
| 355 |
+
"AnchorPplDeltaPercent": -0.05670976370930926,
|
| 356 |
+
"SizeDeltaPercent": 0.9875500745632757,
|
| 357 |
+
"KldRatio": 35.331196581196586,
|
| 358 |
+
"PplAbsRatio": 32.4827044025207
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"Scope": "group",
|
| 362 |
+
"GroupName": "attn_q",
|
| 363 |
+
"RemovedCandidate": "Unsloth-Q3_K_M",
|
| 364 |
+
"AcceptedAnchor": "Unsloth-UD-Q3_K_XL",
|
| 365 |
+
"Reason": "Reason: small size gain (0.04%) but disproportionate damage (KLD x1.58, |PPL| x25.66).",
|
| 366 |
+
"RemovedSizeBytes": 46505957056,
|
| 367 |
+
"RemovedSizeGB": 46.505957056,
|
| 368 |
+
"RemovedSizeGiB": 43.3120476603508,
|
| 369 |
+
"RemovedKld": 0.003666,
|
| 370 |
+
"RemovedPplDeltaPercent": 0.45446277307178795,
|
| 371 |
+
"AnchorSizeBytes": 46522832576,
|
| 372 |
+
"AnchorSizeGB": 46.522832576,
|
| 373 |
+
"AnchorSizeGiB": 43.327764213085175,
|
| 374 |
+
"AnchorKld": 0.002327,
|
| 375 |
+
"AnchorPplDeltaPercent": 0.017708426214895115,
|
| 376 |
+
"SizeDeltaPercent": 0.03627362966868374,
|
| 377 |
+
"KldRatio": 1.5754189944134078,
|
| 378 |
+
"PplAbsRatio": 25.66364551862463
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"Scope": "group",
|
| 382 |
+
"GroupName": "attn_q",
|
| 383 |
+
"RemovedCandidate": "Unsloth-UD-IQ3_XXS",
|
| 384 |
+
"AcceptedAnchor": "Unsloth-UD-Q3_K_XL",
|
| 385 |
+
"Reason": "Reason: small size gain (0.55%) but disproportionate damage (KLD x2.28, |PPL| x4.15).",
|
| 386 |
+
"RemovedSizeBytes": 46267242176,
|
| 387 |
+
"RemovedSizeGB": 46.267242175999996,
|
| 388 |
+
"RemovedSizeGiB": 43.089727103710175,
|
| 389 |
+
"RemovedKld": 0.005316,
|
| 390 |
+
"RemovedPplDeltaPercent": 0.07347302719572618,
|
| 391 |
+
"AnchorSizeBytes": 46522832576,
|
| 392 |
+
"AnchorSizeGB": 46.522832576,
|
| 393 |
+
"AnchorSizeGiB": 43.327764213085175,
|
| 394 |
+
"AnchorKld": 0.002327,
|
| 395 |
+
"AnchorPplDeltaPercent": 0.017708426214895115,
|
| 396 |
+
"SizeDeltaPercent": 0.5493870124577342,
|
| 397 |
+
"KldRatio": 2.284486463257413,
|
| 398 |
+
"PplAbsRatio": 4.149043303121183
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"Scope": "group",
|
| 402 |
+
"GroupName": "attn_q",
|
| 403 |
+
"RemovedCandidate": "Unsloth-UD-IQ2_M",
|
| 404 |
+
"AcceptedAnchor": "Unsloth-UD-Q3_K_XL",
|
| 405 |
+
"Reason": "Reason: small size gain (1.04%) but disproportionate damage (KLD x3.85, |PPL| x28.93).",
|
| 406 |
+
"RemovedSizeBytes": 46040651456,
|
| 407 |
+
"RemovedSizeGB": 46.040651456,
|
| 408 |
+
"RemovedSizeGiB": 42.8786980509758,
|
| 409 |
+
"RemovedKld": 0.008957,
|
| 410 |
+
"RemovedPplDeltaPercent": 0.512349531876957,
|
| 411 |
+
"AnchorSizeBytes": 46522832576,
|
| 412 |
+
"AnchorSizeGB": 46.522832576,
|
| 413 |
+
"AnchorSizeGiB": 43.327764213085175,
|
| 414 |
+
"AnchorKld": 0.002327,
|
| 415 |
+
"AnchorPplDeltaPercent": 0.017708426214895115,
|
| 416 |
+
"SizeDeltaPercent": 1.0364397292712257,
|
| 417 |
+
"KldRatio": 3.849162011173184,
|
| 418 |
+
"PplAbsRatio": 28.93252769385027
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"Scope": "group",
|
| 422 |
+
"GroupName": "attn_q",
|
| 423 |
+
"RemovedCandidate": "Unsloth-UD-Q2_K_XL",
|
| 424 |
+
"AcceptedAnchor": "Unsloth-UD-Q3_K_XL",
|
| 425 |
+
"Reason": "Reason: small size gain (1.04%) but disproportionate damage (KLD x3.85, |PPL| x28.93).",
|
| 426 |
+
"RemovedSizeBytes": 46040651456,
|
| 427 |
+
"RemovedSizeGB": 46.040651456,
|
| 428 |
+
"RemovedSizeGiB": 42.8786980509758,
|
| 429 |
+
"RemovedKld": 0.008957,
|
| 430 |
+
"RemovedPplDeltaPercent": 0.512349531876957,
|
| 431 |
+
"AnchorSizeBytes": 46522832576,
|
| 432 |
+
"AnchorSizeGB": 46.522832576,
|
| 433 |
+
"AnchorSizeGiB": 43.327764213085175,
|
| 434 |
+
"AnchorKld": 0.002327,
|
| 435 |
+
"AnchorPplDeltaPercent": 0.017708426214895115,
|
| 436 |
+
"SizeDeltaPercent": 1.0364397292712257,
|
| 437 |
+
"KldRatio": 3.849162011173184,
|
| 438 |
+
"PplAbsRatio": 28.93252769385027
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"Scope": "group",
|
| 442 |
+
"GroupName": "attn_q",
|
| 443 |
+
"RemovedCandidate": "Unsloth-UD-IQ2_XXS",
|
| 444 |
+
"AcceptedAnchor": "Unsloth-UD-Q3_K_XL",
|
| 445 |
+
"Reason": "Reason: small size gain (1.63%) but disproportionate damage (KLD x9.84, |PPL| x107.94).",
|
| 446 |
+
"RemovedSizeBytes": 45763925696,
|
| 447 |
+
"RemovedSizeGB": 45.76392569600001,
|
| 448 |
+
"RemovedSizeGiB": 42.620977103710175,
|
| 449 |
+
"RemovedKld": 0.022907,
|
| 450 |
+
"RemovedPplDeltaPercent": 1.9115113687026275,
|
| 451 |
+
"AnchorSizeBytes": 46522832576,
|
| 452 |
+
"AnchorSizeGB": 46.522832576,
|
| 453 |
+
"AnchorSizeGiB": 43.327764213085175,
|
| 454 |
+
"AnchorKld": 0.002327,
|
| 455 |
+
"AnchorPplDeltaPercent": 0.017708426214895115,
|
| 456 |
+
"SizeDeltaPercent": 1.6312568216052727,
|
| 457 |
+
"KldRatio": 9.844005156854319,
|
| 458 |
+
"PplAbsRatio": 107.94360523662995
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"Scope": "group",
|
| 462 |
+
"GroupName": "attn_kv",
|
| 463 |
+
"RemovedCandidate": "Unsloth-Q5_K_M",
|
| 464 |
+
"AcceptedAnchor": "Unsloth-UD-Q4_K_XL",
|
| 465 |
+
"Reason": "Reason: small size gain (0.01%) but disproportionate damage (KLD x0.96, |PPL| x4.79).",
|
| 466 |
+
"RemovedSizeBytes": 51562264256,
|
| 467 |
+
"RemovedSizeGB": 51.56226425599999,
|
| 468 |
+
"RemovedSizeGiB": 48.02110069990158,
|
| 469 |
+
"RemovedKld": 0.00043,
|
| 470 |
+
"RemovedPplDeltaPercent": 0.035648684797140257,
|
| 471 |
+
"AnchorSizeBytes": 51567834816,
|
| 472 |
+
"AnchorSizeGB": 51.567834816,
|
| 473 |
+
"AnchorSizeGiB": 48.02628868818283,
|
| 474 |
+
"AnchorKld": 0.00045,
|
| 475 |
+
"AnchorPplDeltaPercent": -0.007436469014706511,
|
| 476 |
+
"SizeDeltaPercent": 0.010802392654018542,
|
| 477 |
+
"KldRatio": 0.9555555555555556,
|
| 478 |
+
"PplAbsRatio": 4.7937649880125495
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"Scope": "group",
|
| 482 |
+
"GroupName": "attn_kv",
|
| 483 |
+
"RemovedCandidate": "Unsloth-UD-Q3_K_XL",
|
| 484 |
+
"AcceptedAnchor": "Unsloth-Q4_K_M",
|
| 485 |
+
"Reason": "Reason: small size gain (0.01%) but disproportionate damage (KLD x0.91, |PPL| x18.69).",
|
| 486 |
+
"RemovedSizeBytes": 51542070976,
|
| 487 |
+
"RemovedSizeGB": 51.542070976000005,
|
| 488 |
+
"RemovedSizeGiB": 48.00229424238205,
|
| 489 |
+
"RemovedKld": 0.000657,
|
| 490 |
+
"RemovedPplDeltaPercent": 0.07267053053946096,
|
| 491 |
+
"AnchorSizeBytes": 51545880256,
|
| 492 |
+
"AnchorSizeGB": 51.545880256,
|
| 493 |
+
"AnchorSizeGiB": 48.00584191083908,
|
| 494 |
+
"AnchorKld": 0.000722,
|
| 495 |
+
"AnchorPplDeltaPercent": -0.003887650468127928,
|
| 496 |
+
"SizeDeltaPercent": 0.007390076531977734,
|
| 497 |
+
"KldRatio": 0.9099722991689752,
|
| 498 |
+
"PplAbsRatio": 18.692660550436514
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"Scope": "group",
|
| 502 |
+
"GroupName": "attn_kv",
|
| 503 |
+
"RemovedCandidate": "Unsloth-UD-IQ3_XXS",
|
| 504 |
+
"AcceptedAnchor": "Unsloth-Q4_K_S",
|
| 505 |
+
"Reason": "Reason: small size gain (0.01%) but disproportionate damage (KLD x1.49, |PPL| x19.97).",
|
| 506 |
+
"RemovedSizeBytes": 51528144576,
|
| 507 |
+
"RemovedSizeGB": 51.528144575999995,
|
| 508 |
+
"RemovedSizeGiB": 47.989324271678925,
|
| 509 |
+
"RemovedKld": 0.001308,
|
| 510 |
+
"RemovedPplDeltaPercent": 0.1923138653588966,
|
| 511 |
+
"AnchorSizeBytes": 51534370496,
|
| 512 |
+
"AnchorSizeGB": 51.534370495999994,
|
| 513 |
+
"AnchorSizeGiB": 47.995122611522675,
|
| 514 |
+
"AnchorKld": 0.000877,
|
| 515 |
+
"AnchorPplDeltaPercent": -0.009629959875166919,
|
| 516 |
+
"SizeDeltaPercent": 0.012081102262582685,
|
| 517 |
+
"KldRatio": 1.4914481185860888,
|
| 518 |
+
"PplAbsRatio": 19.970370370371164
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"Scope": "group",
|
| 522 |
+
"GroupName": "attn_kv",
|
| 523 |
+
"RemovedCandidate": "Unsloth-Q3_K_M",
|
| 524 |
+
"AcceptedAnchor": "Unsloth-Q4_K_S",
|
| 525 |
+
"Reason": "Reason: small size gain (0.02%) but disproportionate damage (KLD x1.70, |PPL| x7.98).",
|
| 526 |
+
"RemovedSizeBytes": 51521877696,
|
| 527 |
+
"RemovedSizeGB": 51.521877696000004,
|
| 528 |
+
"RemovedSizeGiB": 47.98348778486252,
|
| 529 |
+
"RemovedKld": 0.001491,
|
| 530 |
+
"RemovedPplDeltaPercent": 0.07684351315203064,
|
| 531 |
+
"AnchorSizeBytes": 51534370496,
|
| 532 |
+
"AnchorSizeGB": 51.534370495999994,
|
| 533 |
+
"AnchorSizeGiB": 47.995122611522675,
|
| 534 |
+
"AnchorKld": 0.000877,
|
| 535 |
+
"AnchorPplDeltaPercent": -0.009629959875166919,
|
| 536 |
+
"SizeDeltaPercent": 0.024241685461103415,
|
| 537 |
+
"KldRatio": 1.700114025085519,
|
| 538 |
+
"PplAbsRatio": 7.979629629630071
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"Scope": "group",
|
| 542 |
+
"GroupName": "attn_kv",
|
| 543 |
+
"RemovedCandidate": "Unsloth-UD-IQ2_M",
|
| 544 |
+
"AcceptedAnchor": "Unsloth-Q4_K_S",
|
| 545 |
+
"Reason": "Reason: small size gain (0.03%) but disproportionate damage (KLD x1.74, |PPL| x7.10).",
|
| 546 |
+
"RemovedSizeBytes": 51519092416,
|
| 547 |
+
"RemovedSizeGB": 51.519092416,
|
| 548 |
+
"RemovedSizeGiB": 47.98089379072189,
|
| 549 |
+
"RemovedKld": 0.001526,
|
| 550 |
+
"RemovedPplDeltaPercent": 0.06835488185466246,
|
| 551 |
+
"AnchorSizeBytes": 51534370496,
|
| 552 |
+
"AnchorSizeGB": 51.534370495999994,
|
| 553 |
+
"AnchorSizeGiB": 47.995122611522675,
|
| 554 |
+
"AnchorKld": 0.000877,
|
| 555 |
+
"AnchorPplDeltaPercent": -0.009629959875166919,
|
| 556 |
+
"SizeDeltaPercent": 0.029646389104890404,
|
| 557 |
+
"KldRatio": 1.7400228050171038,
|
| 558 |
+
"PplAbsRatio": 7.098148148148711
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"Scope": "group",
|
| 562 |
+
"GroupName": "attn_kv",
|
| 563 |
+
"RemovedCandidate": "Unsloth-UD-Q2_K_XL",
|
| 564 |
+
"AcceptedAnchor": "Unsloth-Q4_K_S",
|
| 565 |
+
"Reason": "Reason: small size gain (0.03%) but disproportionate damage (KLD x1.74, |PPL| x7.10).",
|
| 566 |
+
"RemovedSizeBytes": 51519092416,
|
| 567 |
+
"RemovedSizeGB": 51.519092416,
|
| 568 |
+
"RemovedSizeGiB": 47.98089379072189,
|
| 569 |
+
"RemovedKld": 0.001526,
|
| 570 |
+
"RemovedPplDeltaPercent": 0.06835488185466246,
|
| 571 |
+
"AnchorSizeBytes": 51534370496,
|
| 572 |
+
"AnchorSizeGB": 51.534370495999994,
|
| 573 |
+
"AnchorSizeGiB": 47.995122611522675,
|
| 574 |
+
"AnchorKld": 0.000877,
|
| 575 |
+
"AnchorPplDeltaPercent": -0.009629959875166919,
|
| 576 |
+
"SizeDeltaPercent": 0.029646389104890404,
|
| 577 |
+
"KldRatio": 1.7400228050171038,
|
| 578 |
+
"PplAbsRatio": 7.098148148148711
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"Scope": "group",
|
| 582 |
+
"GroupName": "attn_kv",
|
| 583 |
+
"RemovedCandidate": "Unsloth-UD-IQ2_XXS",
|
| 584 |
+
"AcceptedAnchor": "Unsloth-Q4_K_S",
|
| 585 |
+
"Reason": "Reason: small size gain (0.04%) but disproportionate damage (KLD x2.43, |PPL| x14.11).",
|
| 586 |
+
"RemovedSizeBytes": 51515692736,
|
| 587 |
+
"RemovedSizeGB": 51.515692736000005,
|
| 588 |
+
"RemovedSizeGiB": 47.977727591991425,
|
| 589 |
+
"RemovedKld": 0.002135,
|
| 590 |
+
"RemovedPplDeltaPercent": 0.1359072670530597,
|
| 591 |
+
"AnchorSizeBytes": 51534370496,
|
| 592 |
+
"AnchorSizeGB": 51.534370495999994,
|
| 593 |
+
"AnchorSizeGiB": 47.995122611522675,
|
| 594 |
+
"AnchorKld": 0.000877,
|
| 595 |
+
"AnchorPplDeltaPercent": -0.009629959875166919,
|
| 596 |
+
"SizeDeltaPercent": 0.03624330678774806,
|
| 597 |
+
"KldRatio": 2.434435575826682,
|
| 598 |
+
"PplAbsRatio": 14.112962962963953
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"Scope": "group",
|
| 602 |
+
"GroupName": "attn_kv",
|
| 603 |
+
"RemovedCandidate": "Unsloth-Q3_K_S",
|
| 604 |
+
"AcceptedAnchor": "Unsloth-Q4_K_S",
|
| 605 |
+
"Reason": "Reason: small size gain (0.05%) but disproportionate damage (KLD x7.36, |PPL| x87.26).",
|
| 606 |
+
"RemovedSizeBytes": 51510040256,
|
| 607 |
+
"RemovedSizeGB": 51.510040256,
|
| 608 |
+
"RemovedSizeGiB": 47.97246330976486,
|
| 609 |
+
"RemovedKld": 0.006459,
|
| 610 |
+
"RemovedPplDeltaPercent": 0.8403566651805623,
|
| 611 |
+
"AnchorSizeBytes": 51534370496,
|
| 612 |
+
"AnchorSizeGB": 51.534370495999994,
|
| 613 |
+
"AnchorSizeGiB": 47.995122611522675,
|
| 614 |
+
"AnchorKld": 0.000877,
|
| 615 |
+
"AnchorPplDeltaPercent": -0.009629959875166919,
|
| 616 |
+
"SizeDeltaPercent": 0.04721167594719813,
|
| 617 |
+
"KldRatio": 7.364880273660206,
|
| 618 |
+
"PplAbsRatio": 87.2648148148173
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"Scope": "group",
|
| 622 |
+
"GroupName": "attn_output",
|
| 623 |
+
"RemovedCandidate": "Unsloth-Q5_K_M",
|
| 624 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 625 |
+
"Reason": "Reason: small size gain (0.14%) but disproportionate damage (KLD x1.78, |PPL| x5.88).",
|
| 626 |
+
"RemovedSizeBytes": 51088070336,
|
| 627 |
+
"RemovedSizeGB": 51.088070336,
|
| 628 |
+
"RemovedSizeGiB": 47.579473197460175,
|
| 629 |
+
"RemovedKld": 0.000445,
|
| 630 |
+
"RemovedPplDeltaPercent": 0.045920641997328863,
|
| 631 |
+
"AnchorSizeBytes": 51159094976,
|
| 632 |
+
"AnchorSizeGB": 51.159094976,
|
| 633 |
+
"AnchorSizeGiB": 47.64562004804611,
|
| 634 |
+
"AnchorKld": 0.00025,
|
| 635 |
+
"AnchorPplDeltaPercent": 0.007810967454301176,
|
| 636 |
+
"SizeDeltaPercent": 0.1388309156628346,
|
| 637 |
+
"KldRatio": 1.7799999999999998,
|
| 638 |
+
"PplAbsRatio": 5.878995433791274
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"Scope": "group",
|
| 642 |
+
"GroupName": "attn_output",
|
| 643 |
+
"RemovedCandidate": "Unsloth-Q5_K_S",
|
| 644 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 645 |
+
"Reason": "Reason: small size gain (0.14%) but disproportionate damage (KLD x1.78, |PPL| x5.88).",
|
| 646 |
+
"RemovedSizeBytes": 51088070336,
|
| 647 |
+
"RemovedSizeGB": 51.088070336,
|
| 648 |
+
"RemovedSizeGiB": 47.579473197460175,
|
| 649 |
+
"RemovedKld": 0.000445,
|
| 650 |
+
"RemovedPplDeltaPercent": 0.045920641997328863,
|
| 651 |
+
"AnchorSizeBytes": 51159094976,
|
| 652 |
+
"AnchorSizeGB": 51.159094976,
|
| 653 |
+
"AnchorSizeGiB": 47.64562004804611,
|
| 654 |
+
"AnchorKld": 0.00025,
|
| 655 |
+
"AnchorPplDeltaPercent": 0.007810967454301176,
|
| 656 |
+
"SizeDeltaPercent": 0.1388309156628346,
|
| 657 |
+
"KldRatio": 1.7799999999999998,
|
| 658 |
+
"PplAbsRatio": 5.878995433791274
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"Scope": "group",
|
| 662 |
+
"GroupName": "attn_output",
|
| 663 |
+
"RemovedCandidate": "Unsloth-UD-Q4_K_XL",
|
| 664 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 665 |
+
"Reason": "Reason: small size gain (0.14%) but disproportionate damage (KLD x1.78, |PPL| x5.88).",
|
| 666 |
+
"RemovedSizeBytes": 51088070336,
|
| 667 |
+
"RemovedSizeGB": 51.088070336,
|
| 668 |
+
"RemovedSizeGiB": 47.579473197460175,
|
| 669 |
+
"RemovedKld": 0.000445,
|
| 670 |
+
"RemovedPplDeltaPercent": 0.045920641997328863,
|
| 671 |
+
"AnchorSizeBytes": 51159094976,
|
| 672 |
+
"AnchorSizeGB": 51.159094976,
|
| 673 |
+
"AnchorSizeGiB": 47.64562004804611,
|
| 674 |
+
"AnchorKld": 0.00025,
|
| 675 |
+
"AnchorPplDeltaPercent": 0.007810967454301176,
|
| 676 |
+
"SizeDeltaPercent": 0.1388309156628346,
|
| 677 |
+
"KldRatio": 1.7799999999999998,
|
| 678 |
+
"PplAbsRatio": 5.878995433791274
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"Scope": "group",
|
| 682 |
+
"GroupName": "attn_output",
|
| 683 |
+
"RemovedCandidate": "Unsloth-Q4_K_M",
|
| 684 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 685 |
+
"Reason": "Reason: small size gain (0.27%) but disproportionate damage (KLD x4.03, |PPL| x8.89).",
|
| 686 |
+
"RemovedSizeBytes": 51021223616,
|
| 687 |
+
"RemovedSizeGB": 51.02122361599999,
|
| 688 |
+
"RemovedSizeGiB": 47.517217338085175,
|
| 689 |
+
"RemovedKld": 0.001007,
|
| 690 |
+
"RemovedPplDeltaPercent": -0.06947837717343061,
|
| 691 |
+
"AnchorSizeBytes": 51159094976,
|
| 692 |
+
"AnchorSizeGB": 51.159094976,
|
| 693 |
+
"AnchorSizeGiB": 47.64562004804611,
|
| 694 |
+
"AnchorKld": 0.00025,
|
| 695 |
+
"AnchorPplDeltaPercent": 0.007810967454301176,
|
| 696 |
+
"SizeDeltaPercent": 0.2694953068749142,
|
| 697 |
+
"KldRatio": 4.0280000000000005,
|
| 698 |
+
"PplAbsRatio": 8.8949771689513
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"Scope": "group",
|
| 702 |
+
"GroupName": "attn_output",
|
| 703 |
+
"RemovedCandidate": "Unsloth-Q4_K_S",
|
| 704 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 705 |
+
"Reason": "Reason: small size gain (0.27%) but disproportionate damage (KLD x4.03, |PPL| x8.89).",
|
| 706 |
+
"RemovedSizeBytes": 51021223616,
|
| 707 |
+
"RemovedSizeGB": 51.02122361599999,
|
| 708 |
+
"RemovedSizeGiB": 47.517217338085175,
|
| 709 |
+
"RemovedKld": 0.001007,
|
| 710 |
+
"RemovedPplDeltaPercent": -0.06947837717343061,
|
| 711 |
+
"AnchorSizeBytes": 51159094976,
|
| 712 |
+
"AnchorSizeGB": 51.159094976,
|
| 713 |
+
"AnchorSizeGiB": 47.64562004804611,
|
| 714 |
+
"AnchorKld": 0.00025,
|
| 715 |
+
"AnchorPplDeltaPercent": 0.007810967454301176,
|
| 716 |
+
"SizeDeltaPercent": 0.2694953068749142,
|
| 717 |
+
"KldRatio": 4.0280000000000005,
|
| 718 |
+
"PplAbsRatio": 8.8949771689513
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"Scope": "group",
|
| 722 |
+
"GroupName": "attn_output",
|
| 723 |
+
"RemovedCandidate": "Unsloth-Q3_K_M",
|
| 724 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 725 |
+
"Reason": "Reason: small size gain (0.27%) but disproportionate damage (KLD x4.03, |PPL| x8.89).",
|
| 726 |
+
"RemovedSizeBytes": 51021223616,
|
| 727 |
+
"RemovedSizeGB": 51.02122361599999,
|
| 728 |
+
"RemovedSizeGiB": 47.517217338085175,
|
| 729 |
+
"RemovedKld": 0.001007,
|
| 730 |
+
"RemovedPplDeltaPercent": -0.06947837717343061,
|
| 731 |
+
"AnchorSizeBytes": 51159094976,
|
| 732 |
+
"AnchorSizeGB": 51.159094976,
|
| 733 |
+
"AnchorSizeGiB": 47.64562004804611,
|
| 734 |
+
"AnchorKld": 0.00025,
|
| 735 |
+
"AnchorPplDeltaPercent": 0.007810967454301176,
|
| 736 |
+
"SizeDeltaPercent": 0.2694953068749142,
|
| 737 |
+
"KldRatio": 4.0280000000000005,
|
| 738 |
+
"PplAbsRatio": 8.8949771689513
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"Scope": "group",
|
| 742 |
+
"GroupName": "attn_output",
|
| 743 |
+
"RemovedCandidate": "Unsloth-UD-Q3_K_XL",
|
| 744 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 745 |
+
"Reason": "Reason: small size gain (0.30%) but disproportionate damage (KLD x4.44, |PPL| x21.45).",
|
| 746 |
+
"RemovedSizeBytes": 51004511936,
|
| 747 |
+
"RemovedSizeGB": 51.00451193599999,
|
| 748 |
+
"RemovedSizeGiB": 47.501653373241425,
|
| 749 |
+
"RemovedKld": 0.001109,
|
| 750 |
+
"RemovedPplDeltaPercent": 0.16752563530985906,
|
| 751 |
+
"AnchorSizeBytes": 51159094976,
|
| 752 |
+
"AnchorSizeGB": 51.159094976,
|
| 753 |
+
"AnchorSizeGiB": 47.64562004804611,
|
| 754 |
+
"AnchorKld": 0.00025,
|
| 755 |
+
"AnchorPplDeltaPercent": 0.007810967454301176,
|
| 756 |
+
"SizeDeltaPercent": 0.30216140467793406,
|
| 757 |
+
"KldRatio": 4.436,
|
| 758 |
+
"PplAbsRatio": 21.44748858447869
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"Scope": "group",
|
| 762 |
+
"GroupName": "attn_output",
|
| 763 |
+
"RemovedCandidate": "Unsloth-UD-IQ3_XXS",
|
| 764 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 765 |
+
"Reason": "Reason: small size gain (0.41%) but disproportionate damage (KLD x11.46, |PPL| x40.54).",
|
| 766 |
+
"RemovedSizeBytes": 50950198976,
|
| 767 |
+
"RemovedSizeGB": 50.95019897600001,
|
| 768 |
+
"RemovedSizeGiB": 47.45107048749924,
|
| 769 |
+
"RemovedKld": 0.002865,
|
| 770 |
+
"RemovedPplDeltaPercent": 0.3166473473027254,
|
| 771 |
+
"AnchorSizeBytes": 51159094976,
|
| 772 |
+
"AnchorSizeGB": 51.159094976,
|
| 773 |
+
"AnchorSizeGiB": 47.64562004804611,
|
| 774 |
+
"AnchorKld": 0.00025,
|
| 775 |
+
"AnchorPplDeltaPercent": 0.007810967454301176,
|
| 776 |
+
"SizeDeltaPercent": 0.40832622253774875,
|
| 777 |
+
"KldRatio": 11.459999999999999,
|
| 778 |
+
"PplAbsRatio": 40.538812785394576
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"Scope": "group",
|
| 782 |
+
"GroupName": "attn_output",
|
| 783 |
+
"RemovedCandidate": "Unsloth-UD-IQ2_M",
|
| 784 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 785 |
+
"Reason": "Reason: small size gain (0.41%) but disproportionate damage (KLD x11.46, |PPL| x40.54).",
|
| 786 |
+
"RemovedSizeBytes": 50950198976,
|
| 787 |
+
"RemovedSizeGB": 50.95019897600001,
|
| 788 |
+
"RemovedSizeGiB": 47.45107048749924,
|
| 789 |
+
"RemovedKld": 0.002865,
|
| 790 |
+
"RemovedPplDeltaPercent": 0.3166473473027254,
|
| 791 |
+
"AnchorSizeBytes": 51159094976,
|
| 792 |
+
"AnchorSizeGB": 51.159094976,
|
| 793 |
+
"AnchorSizeGiB": 47.64562004804611,
|
| 794 |
+
"AnchorKld": 0.00025,
|
| 795 |
+
"AnchorPplDeltaPercent": 0.007810967454301176,
|
| 796 |
+
"SizeDeltaPercent": 0.40832622253774875,
|
| 797 |
+
"KldRatio": 11.459999999999999,
|
| 798 |
+
"PplAbsRatio": 40.538812785394576
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"Scope": "group",
|
| 802 |
+
"GroupName": "attn_output",
|
| 803 |
+
"RemovedCandidate": "Unsloth-UD-Q2_K_XL",
|
| 804 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 805 |
+
"Reason": "Reason: small size gain (0.41%) but disproportionate damage (KLD x11.46, |PPL| x40.54).",
|
| 806 |
+
"RemovedSizeBytes": 50950198976,
|
| 807 |
+
"RemovedSizeGB": 50.95019897600001,
|
| 808 |
+
"RemovedSizeGiB": 47.45107048749924,
|
| 809 |
+
"RemovedKld": 0.002865,
|
| 810 |
+
"RemovedPplDeltaPercent": 0.3166473473027254,
|
| 811 |
+
"AnchorSizeBytes": 51159094976,
|
| 812 |
+
"AnchorSizeGB": 51.159094976,
|
| 813 |
+
"AnchorSizeGiB": 47.64562004804611,
|
| 814 |
+
"AnchorKld": 0.00025,
|
| 815 |
+
"AnchorPplDeltaPercent": 0.007810967454301176,
|
| 816 |
+
"SizeDeltaPercent": 0.40832622253774875,
|
| 817 |
+
"KldRatio": 11.459999999999999,
|
| 818 |
+
"PplAbsRatio": 40.538812785394576
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"Scope": "group",
|
| 822 |
+
"GroupName": "attn_output",
|
| 823 |
+
"RemovedCandidate": "Unsloth-UD-IQ2_XXS",
|
| 824 |
+
"AcceptedAnchor": "Unsloth-Q6_K",
|
| 825 |
+
"Reason": "Reason: small size gain (0.51%) but disproportionate damage (KLD x38.04, |PPL| x141.05).",
|
| 826 |
+
"RemovedSizeBytes": 50898343616,
|
| 827 |
+
"RemovedSizeGB": 50.898343616,
|
| 828 |
+
"RemovedSizeGiB": 47.402776420116425,
|
| 829 |
+
"RemovedKld": 0.00951,
|
| 830 |
+
"RemovedPplDeltaPercent": 1.1017209094962188,
|
| 831 |
+
"AnchorSizeBytes": 51159094976,
|
| 832 |
+
"AnchorSizeGB": 51.159094976,
|
| 833 |
+
"AnchorSizeGiB": 47.64562004804611,
|
| 834 |
+
"AnchorKld": 0.00025,
|
| 835 |
+
"AnchorPplDeltaPercent": 0.007810967454301176,
|
| 836 |
+
"SizeDeltaPercent": 0.5096872024853546,
|
| 837 |
+
"KldRatio": 38.04,
|
| 838 |
+
"PplAbsRatio": 141.04794520550035
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"Scope": "group",
|
| 842 |
+
"GroupName": "ffn_up_gate",
|
| 843 |
+
"RemovedCandidate": "Unsloth-IQ4_XS",
|
| 844 |
+
"AcceptedAnchor": "Unsloth-Q4_K_M",
|
| 845 |
+
"Reason": "Reason: small size gain (1.01%) but disproportionate damage (KLD x1.06, |PPL| x4.12).",
|
| 846 |
+
"RemovedSizeBytes": 34777470656,
|
| 847 |
+
"RemovedSizeGB": 34.777470656000006,
|
| 848 |
+
"RemovedSizeGiB": 32.389043509960175,
|
| 849 |
+
"RemovedKld": 0.006919,
|
| 850 |
+
"RemovedPplDeltaPercent": 0.45094962104325464,
|
| 851 |
+
"AnchorSizeBytes": 35133986496,
|
| 852 |
+
"AnchorSizeGB": 35.133986496,
|
| 853 |
+
"AnchorSizeGiB": 32.721074759960175,
|
| 854 |
+
"AnchorKld": 0.006524,
|
| 855 |
+
"AnchorPplDeltaPercent": 0.10933571110120172,
|
| 856 |
+
"SizeDeltaPercent": 1.014732102890144,
|
| 857 |
+
"KldRatio": 1.0605456774984672,
|
| 858 |
+
"PplAbsRatio": 4.124449518838848
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"Scope": "group",
|
| 862 |
+
"GroupName": "ffn_down",
|
| 863 |
+
"RemovedCandidate": "Unsloth-Q6_K",
|
| 864 |
+
"AcceptedAnchor": "Q8_0",
|
| 865 |
+
"Reason": "Reason: small size gain (3.03%) but disproportionate damage (KLD x1.82, |PPL| x38.22).",
|
| 866 |
+
"RemovedSizeBytes": 44955580096,
|
| 867 |
+
"RemovedSizeGB": 44.955580096,
|
| 868 |
+
"RemovedSizeGiB": 41.86814659833908,
|
| 869 |
+
"RemovedKld": 0.000572,
|
| 870 |
+
"RemovedPplDeltaPercent": -0.040213999108335195,
|
| 871 |
+
"AnchorSizeBytes": 46358664896,
|
| 872 |
+
"AnchorSizeGB": 46.35866489599999,
|
| 873 |
+
"AnchorSizeGiB": 43.174871146678925,
|
| 874 |
+
"AnchorKld": 0.000314,
|
| 875 |
+
"AnchorPplDeltaPercent": -0.0010521622826616726,
|
| 876 |
+
"SizeDeltaPercent": 3.02658586727562,
|
| 877 |
+
"KldRatio": 1.821656050955414,
|
| 878 |
+
"PplAbsRatio": 38.220338982885
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"Scope": "group",
|
| 882 |
+
"GroupName": "ffn_down",
|
| 883 |
+
"RemovedCandidate": "Unsloth-UD-Q5_K_XL",
|
| 884 |
+
"AcceptedAnchor": "Q8_0",
|
| 885 |
+
"Reason": "Reason: small size gain (3.03%) but disproportionate damage (KLD x1.82, |PPL| x38.22).",
|
| 886 |
+
"RemovedSizeBytes": 44955580096,
|
| 887 |
+
"RemovedSizeGB": 44.955580096,
|
| 888 |
+
"RemovedSizeGiB": 41.86814659833908,
|
| 889 |
+
"RemovedKld": 0.000572,
|
| 890 |
+
"RemovedPplDeltaPercent": -0.040213999108335195,
|
| 891 |
+
"AnchorSizeBytes": 46358664896,
|
| 892 |
+
"AnchorSizeGB": 46.35866489599999,
|
| 893 |
+
"AnchorSizeGiB": 43.174871146678925,
|
| 894 |
+
"AnchorKld": 0.000314,
|
| 895 |
+
"AnchorPplDeltaPercent": -0.0010521622826616726,
|
| 896 |
+
"SizeDeltaPercent": 3.02658586727562,
|
| 897 |
+
"KldRatio": 1.821656050955414,
|
| 898 |
+
"PplAbsRatio": 38.220338982885
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"Scope": "group",
|
| 902 |
+
"GroupName": "ffn_down",
|
| 903 |
+
"RemovedCandidate": "Unsloth-Q5_K_M",
|
| 904 |
+
"AcceptedAnchor": "Q8_0",
|
| 905 |
+
"Reason": "Reason: small size gain (3.84%) but disproportionate damage (KLD x3.08, |PPL| x40.51).",
|
| 906 |
+
"RemovedSizeBytes": 44576782016,
|
| 907 |
+
"RemovedSizeGB": 44.576782016,
|
| 908 |
+
"RemovedSizeGiB": 41.51536339521408,
|
| 909 |
+
"RemovedKld": 0.000966,
|
| 910 |
+
"RemovedPplDeltaPercent": 0.042621489077130884,
|
| 911 |
+
"AnchorSizeBytes": 46358664896,
|
| 912 |
+
"AnchorSizeGB": 46.35866489599999,
|
| 913 |
+
"AnchorSizeGiB": 43.174871146678925,
|
| 914 |
+
"AnchorKld": 0.000314,
|
| 915 |
+
"AnchorPplDeltaPercent": -0.0010521622826616726,
|
| 916 |
+
"SizeDeltaPercent": 3.8436889500537523,
|
| 917 |
+
"KldRatio": 3.0764331210191083,
|
| 918 |
+
"PplAbsRatio": 40.50847457609922
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"Scope": "group",
|
| 922 |
+
"GroupName": "ffn_down",
|
| 923 |
+
"RemovedCandidate": "Unsloth-Q4_K_S",
|
| 924 |
+
"AcceptedAnchor": "Unsloth-Q5_K_S",
|
| 925 |
+
"Reason": "Reason: small size gain (1.44%) but disproportionate damage (KLD x2.61, |PPL| x1.55).",
|
| 926 |
+
"RemovedSizeBytes": 43551102656,
|
| 927 |
+
"RemovedSizeGB": 43.551102656000005,
|
| 928 |
+
"RemovedSizeGiB": 40.560125052928925,
|
| 929 |
+
"RemovedKld": 0.004065,
|
| 930 |
+
"RemovedPplDeltaPercent": 0.10173874275523584,
|
| 931 |
+
"AnchorSizeBytes": 44186146496,
|
| 932 |
+
"AnchorSizeGB": 44.186146496,
|
| 933 |
+
"AnchorSizeGiB": 41.151555716991425,
|
| 934 |
+
"AnchorKld": 0.001558,
|
| 935 |
+
"AnchorPplDeltaPercent": 0.06573339277753149,
|
| 936 |
+
"SizeDeltaPercent": 1.4372012278950101,
|
| 937 |
+
"KldRatio": 2.609114249037227,
|
| 938 |
+
"PplAbsRatio": 1.547748236570735
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"Scope": "group",
|
| 942 |
+
"GroupName": "ffn_down",
|
| 943 |
+
"RemovedCandidate": "Unsloth-Q3_K_M",
|
| 944 |
+
"AcceptedAnchor": "Unsloth-Q5_K_S",
|
| 945 |
+
"Reason": "Reason: small size gain (1.54%) but disproportionate damage (KLD x2.67, |PPL| x3.14).",
|
| 946 |
+
"RemovedSizeBytes": 43506538176,
|
| 947 |
+
"RemovedSizeGB": 43.506538176,
|
| 948 |
+
"RemovedSizeGiB": 40.518621146678925,
|
| 949 |
+
"RemovedKld": 0.004153,
|
| 950 |
+
"RemovedPplDeltaPercent": 0.20650913954525846,
|
| 951 |
+
"AnchorSizeBytes": 44186146496,
|
| 952 |
+
"AnchorSizeGB": 44.186146496,
|
| 953 |
+
"AnchorSizeGiB": 41.151555716991425,
|
| 954 |
+
"AnchorKld": 0.001558,
|
| 955 |
+
"AnchorPplDeltaPercent": 0.06573339277753149,
|
| 956 |
+
"SizeDeltaPercent": 1.5380574544139582,
|
| 957 |
+
"KldRatio": 2.665596919127086,
|
| 958 |
+
"PplAbsRatio": 3.1416169289202722
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"Scope": "group",
|
| 962 |
+
"GroupName": "ffn_down",
|
| 963 |
+
"RemovedCandidate": "Unsloth-IQ4_NL",
|
| 964 |
+
"AcceptedAnchor": "Unsloth-Q5_K_S",
|
| 965 |
+
"Reason": "Reason: small size gain (1.64%) but disproportionate damage (KLD x2.81, |PPL| x4.66).",
|
| 966 |
+
"RemovedSizeBytes": 43461973696,
|
| 967 |
+
"RemovedSizeGB": 43.461973696,
|
| 968 |
+
"RemovedSizeGiB": 40.477117240428925,
|
| 969 |
+
"RemovedKld": 0.004375,
|
| 970 |
+
"RemovedPplDeltaPercent": 0.30619705751226267,
|
| 971 |
+
"AnchorSizeBytes": 44186146496,
|
| 972 |
+
"AnchorSizeGB": 44.186146496,
|
| 973 |
+
"AnchorSizeGiB": 41.151555716991425,
|
| 974 |
+
"AnchorKld": 0.001558,
|
| 975 |
+
"AnchorPplDeltaPercent": 0.06573339277753149,
|
| 976 |
+
"SizeDeltaPercent": 1.6389136809329063,
|
| 977 |
+
"KldRatio": 2.80808729139923,
|
| 978 |
+
"PplAbsRatio": 4.65816603364074
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"Scope": "group",
|
| 982 |
+
"GroupName": "ffn_down",
|
| 983 |
+
"RemovedCandidate": "Unsloth-UD-Q3_K_XL",
|
| 984 |
+
"AcceptedAnchor": "Unsloth-Q5_K_S",
|
| 985 |
+
"Reason": "Reason: small size gain (2.05%) but disproportionate damage (KLD x2.93, |PPL| x6.73).",
|
| 986 |
+
"RemovedSizeBytes": 43280930496,
|
| 987 |
+
"RemovedSizeGB": 43.280930496,
|
| 988 |
+
"RemovedSizeGiB": 40.3085076212883,
|
| 989 |
+
"RemovedKld": 0.004566,
|
| 990 |
+
"RemovedPplDeltaPercent": 0.4420864913062918,
|
| 991 |
+
"AnchorSizeBytes": 44186146496,
|
| 992 |
+
"AnchorSizeGB": 44.186146496,
|
| 993 |
+
"AnchorSizeGiB": 41.151555716991425,
|
| 994 |
+
"AnchorKld": 0.001558,
|
| 995 |
+
"AnchorPplDeltaPercent": 0.06573339277753149,
|
| 996 |
+
"SizeDeltaPercent": 2.048642101166133,
|
| 997 |
+
"KldRatio": 2.9306803594351734,
|
| 998 |
+
"PplAbsRatio": 6.725447639717793
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"Scope": "group",
|
| 1002 |
+
"GroupName": "ffn_down",
|
| 1003 |
+
"RemovedCandidate": "Unsloth-UD-IQ3_XXS",
|
| 1004 |
+
"AcceptedAnchor": "Unsloth-Q5_K_S",
|
| 1005 |
+
"Reason": "Reason: small size gain (3.38%) but disproportionate damage (KLD x8.28, |PPL| x16.29).",
|
| 1006 |
+
"RemovedSizeBytes": 42692540096,
|
| 1007 |
+
"RemovedSizeGB": 42.692540096,
|
| 1008 |
+
"RemovedSizeGiB": 39.76052635908127,
|
| 1009 |
+
"RemovedKld": 0.012893,
|
| 1010 |
+
"RemovedPplDeltaPercent": 1.0709228711546994,
|
| 1011 |
+
"AnchorSizeBytes": 44186146496,
|
| 1012 |
+
"AnchorSizeGB": 44.186146496,
|
| 1013 |
+
"AnchorSizeGiB": 41.151555716991425,
|
| 1014 |
+
"AnchorKld": 0.001558,
|
| 1015 |
+
"AnchorPplDeltaPercent": 0.06573339277753149,
|
| 1016 |
+
"SizeDeltaPercent": 3.3802594669241195,
|
| 1017 |
+
"KldRatio": 8.275353016688062,
|
| 1018 |
+
"PplAbsRatio": 16.291915355398398
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"Scope": "group",
|
| 1022 |
+
"GroupName": "ffn_down",
|
| 1023 |
+
"RemovedCandidate": "Unsloth-Q3_K_S",
|
| 1024 |
+
"AcceptedAnchor": "Unsloth-Q5_K_S",
|
| 1025 |
+
"Reason": "Reason: small size gain (3.38%) but disproportionate damage (KLD x11.23, |PPL| x10.28).",
|
| 1026 |
+
"RemovedSizeBytes": 42692540096,
|
| 1027 |
+
"RemovedSizeGB": 42.692540096,
|
| 1028 |
+
"RemovedSizeGiB": 39.76052635908127,
|
| 1029 |
+
"RemovedKld": 0.0175,
|
| 1030 |
+
"RemovedPplDeltaPercent": 0.6760410164957641,
|
| 1031 |
+
"AnchorSizeBytes": 44186146496,
|
| 1032 |
+
"AnchorSizeGB": 44.186146496,
|
| 1033 |
+
"AnchorSizeGiB": 41.151555716991425,
|
| 1034 |
+
"AnchorKld": 0.001558,
|
| 1035 |
+
"AnchorPplDeltaPercent": 0.06573339277753149,
|
| 1036 |
+
"SizeDeltaPercent": 3.3802594669241195,
|
| 1037 |
+
"KldRatio": 11.23234916559692,
|
| 1038 |
+
"PplAbsRatio": 10.284590341833741
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"Scope": "group",
|
| 1042 |
+
"GroupName": "ffn_down",
|
| 1043 |
+
"RemovedCandidate": "Unsloth-UD-IQ2_M",
|
| 1044 |
+
"AcceptedAnchor": "Unsloth-Q5_K_S",
|
| 1045 |
+
"Reason": "Reason: small size gain (3.96%) but disproportionate damage (KLD x14.85, |PPL| x13.57).",
|
| 1046 |
+
"RemovedSizeBytes": 42434205376,
|
| 1047 |
+
"RemovedSizeGB": 42.43420537600001,
|
| 1048 |
+
"RemovedSizeGiB": 39.5199334025383,
|
| 1049 |
+
"RemovedKld": 0.023144,
|
| 1050 |
+
"RemovedPplDeltaPercent": 0.8921444493981283,
|
| 1051 |
+
"AnchorSizeBytes": 44186146496,
|
| 1052 |
+
"AnchorSizeGB": 44.186146496,
|
| 1053 |
+
"AnchorSizeGiB": 41.151555716991425,
|
| 1054 |
+
"AnchorKld": 0.001558,
|
| 1055 |
+
"AnchorPplDeltaPercent": 0.06573339277753149,
|
| 1056 |
+
"SizeDeltaPercent": 3.9649104050261466,
|
| 1057 |
+
"KldRatio": 14.854942233632864,
|
| 1058 |
+
"PplAbsRatio": 13.572164948453333
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"Scope": "group",
|
| 1062 |
+
"GroupName": "ffn_down",
|
| 1063 |
+
"RemovedCandidate": "Unsloth-UD-Q2_K_XL",
|
| 1064 |
+
"AcceptedAnchor": "Unsloth-Q5_K_S",
|
| 1065 |
+
"Reason": "Reason: small size gain (3.96%) but disproportionate damage (KLD x14.85, |PPL| x13.57).",
|
| 1066 |
+
"RemovedSizeBytes": 42434205376,
|
| 1067 |
+
"RemovedSizeGB": 42.43420537600001,
|
| 1068 |
+
"RemovedSizeGiB": 39.5199334025383,
|
| 1069 |
+
"RemovedKld": 0.023144,
|
| 1070 |
+
"RemovedPplDeltaPercent": 0.8921444493981283,
|
| 1071 |
+
"AnchorSizeBytes": 44186146496,
|
| 1072 |
+
"AnchorSizeGB": 44.186146496,
|
| 1073 |
+
"AnchorSizeGiB": 41.151555716991425,
|
| 1074 |
+
"AnchorKld": 0.001558,
|
| 1075 |
+
"AnchorPplDeltaPercent": 0.06573339277753149,
|
| 1076 |
+
"SizeDeltaPercent": 3.9649104050261466,
|
| 1077 |
+
"KldRatio": 14.854942233632864,
|
| 1078 |
+
"PplAbsRatio": 13.572164948453333
|
| 1079 |
+
}
|
| 1080 |
+
],
|
| 1081 |
+
"synergySecondChances": [
|
| 1082 |
+
{
|
| 1083 |
+
"SynergyName": "ffn_up_gate\u002Bffn_down",
|
| 1084 |
+
"GroupName": "ffn_down",
|
| 1085 |
+
"PeerGroupName": "ffn_up_gate",
|
| 1086 |
+
"RestoredCandidate": "Unsloth-UD-Q5_K_XL",
|
| 1087 |
+
"AcceptedAnchor": "Q8_0",
|
| 1088 |
+
"OriginalBadTradeReason": "Reason: small size gain (3.03%) but disproportionate damage (KLD x1.82, |PPL| x38.22).",
|
| 1089 |
+
"SecondChanceReason": "synergy \u0027ffn_up_gate\u002Bffn_down\u0027 second chance because \u0027Unsloth-UD-Q5_K_XL\u0027 survived in peer group \u0027ffn_up_gate\u0027 and the original bad-trade decision does not survive KLD-only review vs anchor \u0027Q8_0\u0027",
|
| 1090 |
+
"RestoredSizeBytes": 44955580096,
|
| 1091 |
+
"RestoredKld": 0.000572,
|
| 1092 |
+
"RestoredPplDeltaPercent": -0.040213999108335195,
|
| 1093 |
+
"AnchorSizeBytes": 46358664896,
|
| 1094 |
+
"AnchorKld": 0.000314,
|
| 1095 |
+
"AnchorPplDeltaPercent": -0.0010521622826616726
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"SynergyName": "ffn_up_gate\u002Bffn_down",
|
| 1099 |
+
"GroupName": "ffn_down",
|
| 1100 |
+
"PeerGroupName": "ffn_up_gate",
|
| 1101 |
+
"RestoredCandidate": "Unsloth-Q6_K",
|
| 1102 |
+
"AcceptedAnchor": "Q8_0",
|
| 1103 |
+
"OriginalBadTradeReason": "Reason: small size gain (3.03%) but disproportionate damage (KLD x1.82, |PPL| x38.22).",
|
| 1104 |
+
"SecondChanceReason": "synergy \u0027ffn_up_gate\u002Bffn_down\u0027 second chance because \u0027Unsloth-Q6_K\u0027 survived in peer group \u0027ffn_up_gate\u0027 and the original bad-trade decision does not survive KLD-only review vs anchor \u0027Q8_0\u0027",
|
| 1105 |
+
"RestoredSizeBytes": 44955580096,
|
| 1106 |
+
"RestoredKld": 0.000572,
|
| 1107 |
+
"RestoredPplDeltaPercent": -0.040213999108335195,
|
| 1108 |
+
"AnchorSizeBytes": 46358664896,
|
| 1109 |
+
"AnchorKld": 0.000314,
|
| 1110 |
+
"AnchorPplDeltaPercent": -0.0010521622826616726
|
| 1111 |
+
}
|
| 1112 |
+
],
|
| 1113 |
+
"notes": [
|
| 1114 |
+
"Bad trade elimination: \u0027Unsloth-Q4_K_M\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_M\u0027 for \u0027embeddings\u0027. Reason: small size gain (0.32%) but disproportionate damage (KLD x1.46, |PPL| x6.07).",
|
| 1115 |
+
"Bad trade elimination: \u0027Unsloth-UD-Q4_K_XL\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_M\u0027 for \u0027embeddings\u0027. Reason: small size gain (0.32%) but disproportionate damage (KLD x1.46, |PPL| x6.07).",
|
| 1116 |
+
"Bad trade elimination: \u0027Unsloth-Q4_K_S\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_M\u0027 for \u0027embeddings\u0027. Reason: small size gain (0.32%) but disproportionate damage (KLD x1.46, |PPL| x6.07).",
|
| 1117 |
+
"Bad trade elimination: \u0027Unsloth-IQ4_NL\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_M\u0027 for \u0027embeddings\u0027. Reason: small size gain (0.32%) but disproportionate damage (KLD x1.46, |PPL| x6.07).",
|
| 1118 |
+
"Bad trade elimination: \u0027Unsloth-IQ4_XS\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_M\u0027 for \u0027embeddings\u0027. Reason: small size gain (0.32%) but disproportionate damage (KLD x1.46, |PPL| x6.07).",
|
| 1119 |
+
"Bad trade elimination: \u0027Unsloth-Q3_K_M\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_M\u0027 for \u0027embeddings\u0027. Reason: small size gain (0.65%) but disproportionate damage (KLD x3.19, |PPL| x39.63).",
|
| 1120 |
+
"Bad trade elimination: \u0027Unsloth-Q3_K_S\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_M\u0027 for \u0027embeddings\u0027. Reason: small size gain (0.65%) but disproportionate damage (KLD x3.19, |PPL| x39.63).",
|
| 1121 |
+
"Bad trade elimination: \u0027Unsloth-UD-Q3_K_XL\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_M\u0027 for \u0027embeddings\u0027. Reason: small size gain (0.65%) but disproportionate damage (KLD x3.19, |PPL| x39.63).",
|
| 1122 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ3_XXS\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_M\u0027 for \u0027embeddings\u0027. Reason: small size gain (0.91%) but disproportionate damage (KLD x7.11, |PPL| x1.04).",
|
| 1123 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ2_M\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_M\u0027 for \u0027embeddings\u0027. Reason: small size gain (0.91%) but disproportionate damage (KLD x7.11, |PPL| x1.04).",
|
| 1124 |
+
"Bad trade elimination: \u0027Unsloth-UD-Q2_K_XL\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_M\u0027 for \u0027embeddings\u0027. Reason: small size gain (0.91%) but disproportionate damage (KLD x7.11, |PPL| x1.04).",
|
| 1125 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ2_XXS\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_M\u0027 for \u0027embeddings\u0027. Reason: small size gain (0.91%) but disproportionate damage (KLD x7.11, |PPL| x1.04).",
|
| 1126 |
+
"Bad trade elimination: \u0027Unsloth-UD-Q3_K_XL\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027lm_head\u0027. Reason: small size gain (0.34%) but disproportionate damage (KLD x2.65, |PPL| x2.32).",
|
| 1127 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ3_XXS\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027lm_head\u0027. Reason: small size gain (0.34%) but disproportionate damage (KLD x2.65, |PPL| x2.32).",
|
| 1128 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ2_M\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027lm_head\u0027. Reason: small size gain (0.99%) but disproportionate damage (KLD x35.33, |PPL| x32.48).",
|
| 1129 |
+
"Bad trade elimination: \u0027Unsloth-UD-Q2_K_XL\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027lm_head\u0027. Reason: small size gain (0.99%) but disproportionate damage (KLD x35.33, |PPL| x32.48).",
|
| 1130 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ2_XXS\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027lm_head\u0027. Reason: small size gain (0.99%) but disproportionate damage (KLD x35.33, |PPL| x32.48).",
|
| 1131 |
+
"Bad trade elimination: \u0027Unsloth-Q3_K_M\u0027 removed vs accepted anchor \u0027Unsloth-UD-Q3_K_XL\u0027 for \u0027attn_q\u0027. Reason: small size gain (0.04%) but disproportionate damage (KLD x1.58, |PPL| x25.66).",
|
| 1132 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ3_XXS\u0027 removed vs accepted anchor \u0027Unsloth-UD-Q3_K_XL\u0027 for \u0027attn_q\u0027. Reason: small size gain (0.55%) but disproportionate damage (KLD x2.28, |PPL| x4.15).",
|
| 1133 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ2_M\u0027 removed vs accepted anchor \u0027Unsloth-UD-Q3_K_XL\u0027 for \u0027attn_q\u0027. Reason: small size gain (1.04%) but disproportionate damage (KLD x3.85, |PPL| x28.93).",
|
| 1134 |
+
"Bad trade elimination: \u0027Unsloth-UD-Q2_K_XL\u0027 removed vs accepted anchor \u0027Unsloth-UD-Q3_K_XL\u0027 for \u0027attn_q\u0027. Reason: small size gain (1.04%) but disproportionate damage (KLD x3.85, |PPL| x28.93).",
|
| 1135 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ2_XXS\u0027 removed vs accepted anchor \u0027Unsloth-UD-Q3_K_XL\u0027 for \u0027attn_q\u0027. Reason: small size gain (1.63%) but disproportionate damage (KLD x9.84, |PPL| x107.94).",
|
| 1136 |
+
"Bad trade elimination: \u0027Unsloth-Q5_K_M\u0027 removed vs accepted anchor \u0027Unsloth-UD-Q4_K_XL\u0027 for \u0027attn_kv\u0027. Reason: small size gain (0.01%) but disproportionate damage (KLD x0.96, |PPL| x4.79).",
|
| 1137 |
+
"Bad trade elimination: \u0027Unsloth-UD-Q3_K_XL\u0027 removed vs accepted anchor \u0027Unsloth-Q4_K_M\u0027 for \u0027attn_kv\u0027. Reason: small size gain (0.01%) but disproportionate damage (KLD x0.91, |PPL| x18.69).",
|
| 1138 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ3_XXS\u0027 removed vs accepted anchor \u0027Unsloth-Q4_K_S\u0027 for \u0027attn_kv\u0027. Reason: small size gain (0.01%) but disproportionate damage (KLD x1.49, |PPL| x19.97).",
|
| 1139 |
+
"Bad trade elimination: \u0027Unsloth-Q3_K_M\u0027 removed vs accepted anchor \u0027Unsloth-Q4_K_S\u0027 for \u0027attn_kv\u0027. Reason: small size gain (0.02%) but disproportionate damage (KLD x1.70, |PPL| x7.98).",
|
| 1140 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ2_M\u0027 removed vs accepted anchor \u0027Unsloth-Q4_K_S\u0027 for \u0027attn_kv\u0027. Reason: small size gain (0.03%) but disproportionate damage (KLD x1.74, |PPL| x7.10).",
|
| 1141 |
+
"Bad trade elimination: \u0027Unsloth-UD-Q2_K_XL\u0027 removed vs accepted anchor \u0027Unsloth-Q4_K_S\u0027 for \u0027attn_kv\u0027. Reason: small size gain (0.03%) but disproportionate damage (KLD x1.74, |PPL| x7.10).",
|
| 1142 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ2_XXS\u0027 removed vs accepted anchor \u0027Unsloth-Q4_K_S\u0027 for \u0027attn_kv\u0027. Reason: small size gain (0.04%) but disproportionate damage (KLD x2.43, |PPL| x14.11).",
|
| 1143 |
+
"Bad trade elimination: \u0027Unsloth-Q3_K_S\u0027 removed vs accepted anchor \u0027Unsloth-Q4_K_S\u0027 for \u0027attn_kv\u0027. Reason: small size gain (0.05%) but disproportionate damage (KLD x7.36, |PPL| x87.26).",
|
| 1144 |
+
"Bad trade elimination: \u0027Unsloth-Q5_K_M\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027attn_output\u0027. Reason: small size gain (0.14%) but disproportionate damage (KLD x1.78, |PPL| x5.88).",
|
| 1145 |
+
"Bad trade elimination: \u0027Unsloth-Q5_K_S\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027attn_output\u0027. Reason: small size gain (0.14%) but disproportionate damage (KLD x1.78, |PPL| x5.88).",
|
| 1146 |
+
"Bad trade elimination: \u0027Unsloth-UD-Q4_K_XL\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027attn_output\u0027. Reason: small size gain (0.14%) but disproportionate damage (KLD x1.78, |PPL| x5.88).",
|
| 1147 |
+
"Bad trade elimination: \u0027Unsloth-Q4_K_M\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027attn_output\u0027. Reason: small size gain (0.27%) but disproportionate damage (KLD x4.03, |PPL| x8.89).",
|
| 1148 |
+
"Bad trade elimination: \u0027Unsloth-Q4_K_S\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027attn_output\u0027. Reason: small size gain (0.27%) but disproportionate damage (KLD x4.03, |PPL| x8.89).",
|
| 1149 |
+
"Bad trade elimination: \u0027Unsloth-Q3_K_M\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027attn_output\u0027. Reason: small size gain (0.27%) but disproportionate damage (KLD x4.03, |PPL| x8.89).",
|
| 1150 |
+
"Bad trade elimination: \u0027Unsloth-UD-Q3_K_XL\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027attn_output\u0027. Reason: small size gain (0.30%) but disproportionate damage (KLD x4.44, |PPL| x21.45).",
|
| 1151 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ3_XXS\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027attn_output\u0027. Reason: small size gain (0.41%) but disproportionate damage (KLD x11.46, |PPL| x40.54).",
|
| 1152 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ2_M\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027attn_output\u0027. Reason: small size gain (0.41%) but disproportionate damage (KLD x11.46, |PPL| x40.54).",
|
| 1153 |
+
"Bad trade elimination: \u0027Unsloth-UD-Q2_K_XL\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027attn_output\u0027. Reason: small size gain (0.41%) but disproportionate damage (KLD x11.46, |PPL| x40.54).",
|
| 1154 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ2_XXS\u0027 removed vs accepted anchor \u0027Unsloth-Q6_K\u0027 for \u0027attn_output\u0027. Reason: small size gain (0.51%) but disproportionate damage (KLD x38.04, |PPL| x141.05).",
|
| 1155 |
+
"Bad trade elimination: \u0027Unsloth-IQ4_XS\u0027 removed vs accepted anchor \u0027Unsloth-Q4_K_M\u0027 for \u0027ffn_up_gate\u0027. Reason: small size gain (1.01%) but disproportionate damage (KLD x1.06, |PPL| x4.12).",
|
| 1156 |
+
"Bad trade elimination: \u0027Unsloth-Q6_K\u0027 removed vs accepted anchor \u0027Q8_0\u0027 for \u0027ffn_down\u0027. Reason: small size gain (3.03%) but disproportionate damage (KLD x1.82, |PPL| x38.22).",
|
| 1157 |
+
"Bad trade elimination: \u0027Unsloth-UD-Q5_K_XL\u0027 removed vs accepted anchor \u0027Q8_0\u0027 for \u0027ffn_down\u0027. Reason: small size gain (3.03%) but disproportionate damage (KLD x1.82, |PPL| x38.22).",
|
| 1158 |
+
"Bad trade elimination: \u0027Unsloth-Q5_K_M\u0027 removed vs accepted anchor \u0027Q8_0\u0027 for \u0027ffn_down\u0027. Reason: small size gain (3.84%) but disproportionate damage (KLD x3.08, |PPL| x40.51).",
|
| 1159 |
+
"Bad trade elimination: \u0027Unsloth-Q4_K_S\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_S\u0027 for \u0027ffn_down\u0027. Reason: small size gain (1.44%) but disproportionate damage (KLD x2.61, |PPL| x1.55).",
|
| 1160 |
+
"Bad trade elimination: \u0027Unsloth-Q3_K_M\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_S\u0027 for \u0027ffn_down\u0027. Reason: small size gain (1.54%) but disproportionate damage (KLD x2.67, |PPL| x3.14).",
|
| 1161 |
+
"Bad trade elimination: \u0027Unsloth-IQ4_NL\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_S\u0027 for \u0027ffn_down\u0027. Reason: small size gain (1.64%) but disproportionate damage (KLD x2.81, |PPL| x4.66).",
|
| 1162 |
+
"Bad trade elimination: \u0027Unsloth-UD-Q3_K_XL\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_S\u0027 for \u0027ffn_down\u0027. Reason: small size gain (2.05%) but disproportionate damage (KLD x2.93, |PPL| x6.73).",
|
| 1163 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ3_XXS\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_S\u0027 for \u0027ffn_down\u0027. Reason: small size gain (3.38%) but disproportionate damage (KLD x8.28, |PPL| x16.29).",
|
| 1164 |
+
"Bad trade elimination: \u0027Unsloth-Q3_K_S\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_S\u0027 for \u0027ffn_down\u0027. Reason: small size gain (3.38%) but disproportionate damage (KLD x11.23, |PPL| x10.28).",
|
| 1165 |
+
"Bad trade elimination: \u0027Unsloth-UD-IQ2_M\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_S\u0027 for \u0027ffn_down\u0027. Reason: small size gain (3.96%) but disproportionate damage (KLD x14.85, |PPL| x13.57).",
|
| 1166 |
+
"Bad trade elimination: \u0027Unsloth-UD-Q2_K_XL\u0027 removed vs accepted anchor \u0027Unsloth-Q5_K_S\u0027 for \u0027ffn_down\u0027. Reason: small size gain (3.96%) but disproportionate damage (KLD x14.85, |PPL| x13.57).",
|
| 1167 |
+
"Synergy second-chance rejected: \u0027Unsloth-UD-IQ2_M\u0027 remained removed for \u0027ffn_down\u0027 even though it survived in \u0027ffn_up_gate\u0027 under synergy \u0027ffn_up_gate\u002Bffn_down\u0027. Reason: small size gain (3.96%) but disproportionate KLD damage after PPL was ignored for synergy review (KLD x14.85).",
|
| 1168 |
+
"Synergy second-chance rejected: \u0027Unsloth-UD-Q2_K_XL\u0027 remained removed for \u0027ffn_down\u0027 even though it survived in \u0027ffn_up_gate\u0027 under synergy \u0027ffn_up_gate\u002Bffn_down\u0027. Reason: small size gain (3.96%) but disproportionate KLD damage after PPL was ignored for synergy review (KLD x14.85).",
|
| 1169 |
+
"Synergy second-chance rejected: \u0027Unsloth-UD-IQ3_XXS\u0027 remained removed for \u0027ffn_down\u0027 even though it survived in \u0027ffn_up_gate\u0027 under synergy \u0027ffn_up_gate\u002Bffn_down\u0027. Reason: small size gain (3.38%) but disproportionate KLD damage after PPL was ignored for synergy review (KLD x8.28).",
|
| 1170 |
+
"Synergy second-chance rejected: \u0027Unsloth-UD-Q3_K_XL\u0027 remained removed for \u0027ffn_down\u0027 even though it survived in \u0027ffn_up_gate\u0027 under synergy \u0027ffn_up_gate\u002Bffn_down\u0027. Reason: small size gain (2.05%) but disproportionate KLD damage after PPL was ignored for synergy review (KLD x2.93).",
|
| 1171 |
+
"Synergy second-chance rejected: \u0027Unsloth-Q4_K_S\u0027 remained removed for \u0027ffn_down\u0027 even though it survived in \u0027ffn_up_gate\u0027 under synergy \u0027ffn_up_gate\u002Bffn_down\u0027. Reason: small size gain (1.44%) but disproportionate KLD damage after PPL was ignored for synergy review (KLD x2.61).",
|
| 1172 |
+
"Synergy second-chance rejected: \u0027Unsloth-Q5_K_M\u0027 remained removed for \u0027ffn_down\u0027 even though it survived in \u0027ffn_up_gate\u0027 under synergy \u0027ffn_up_gate\u002Bffn_down\u0027. Reason: small size gain (3.84%) but disproportionate KLD damage after PPL was ignored for synergy review (KLD x3.08).",
|
| 1173 |
+
"Synergy second-chance restored: \u0027Unsloth-UD-Q5_K_XL\u0027 restored for \u0027ffn_down\u0027 because it survived in peer group \u0027ffn_up_gate\u0027 under synergy \u0027ffn_up_gate\u002Bffn_down\u0027, and KLD-only bad-trade review did not eliminate it vs anchor \u0027Q8_0\u0027.",
|
| 1174 |
+
"Synergy second-chance restored: \u0027Unsloth-Q6_K\u0027 restored for \u0027ffn_down\u0027 because it survived in peer group \u0027ffn_up_gate\u0027 under synergy \u0027ffn_up_gate\u002Bffn_down\u0027, and KLD-only bad-trade review did not eliminate it vs anchor \u0027Q8_0\u0027."
|
| 1175 |
+
]
|
| 1176 |
+
}
|
magicquant-manifest/magicquant.clone-configs.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
magicquant-manifest/magicquant.final-survivors.json
ADDED
|
@@ -0,0 +1,495 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"key": "0:0:0:0:0:0:0:0:0:0",
|
| 4 |
+
"fileName": "Qwen3.8-27B-LM-Q8_0.gguf",
|
| 5 |
+
"displayName": "LM-Q8_0",
|
| 6 |
+
"shortName": "LM-Q8_0",
|
| 7 |
+
"provider": "llama.cpp",
|
| 8 |
+
"quantFamily": "Q8_0",
|
| 9 |
+
"isHybrid": false,
|
| 10 |
+
"isExternalPureBaseline": false,
|
| 11 |
+
"isExternalRebuiltBaseline": false,
|
| 12 |
+
"isMaterializedTensorMapped": false,
|
| 13 |
+
"isExternalReference": false,
|
| 14 |
+
"downloadTarget": "./../../resolve/main/Qwen3.8-27B-LM-Q8_0.gguf?download=true",
|
| 15 |
+
"kld": 0.000712,
|
| 16 |
+
"ppl": 5.609837,
|
| 17 |
+
"pplDeltaPercent": 0.041676326348636855,
|
| 18 |
+
"sizeBytes": 29047084736,
|
| 19 |
+
"sizeGB": 29.047084736000002,
|
| 20 |
+
"sizeGiB": 27.052205741405487,
|
| 21 |
+
"expectedSizeBytes": 29047084736,
|
| 22 |
+
"actualSizeBytes": 29047084736,
|
| 23 |
+
"usedImatrix": true,
|
| 24 |
+
"replacedArtifacts": []
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"key": "118:119:119:119:117:119:119:119:0:0",
|
| 28 |
+
"fileName": "Qwen3.8-27B-MQ-Q6_K_1.gguf",
|
| 29 |
+
"displayName": "MQ-Q6_K_1",
|
| 30 |
+
"shortName": "MQ-Q6_K_1",
|
| 31 |
+
"provider": "MagicQuant",
|
| 32 |
+
"quantFamily": "Q6_K",
|
| 33 |
+
"isHybrid": true,
|
| 34 |
+
"isExternalPureBaseline": false,
|
| 35 |
+
"isExternalRebuiltBaseline": false,
|
| 36 |
+
"isMaterializedTensorMapped": true,
|
| 37 |
+
"isExternalReference": false,
|
| 38 |
+
"downloadTarget": "./../../resolve/main/Qwen3.8-27B-MQ-Q6_K_1.gguf?download=true",
|
| 39 |
+
"kld": 0.001052,
|
| 40 |
+
"ppl": 5.611099,
|
| 41 |
+
"pplDeltaPercent": 0.06418189924209278,
|
| 42 |
+
"sizeBytes": 25941640896,
|
| 43 |
+
"sizeGB": 25.941640896,
|
| 44 |
+
"sizeGiB": 24.16003578901291,
|
| 45 |
+
"expectedSizeBytes": 25941640896,
|
| 46 |
+
"actualSizeBytes": 25941640896,
|
| 47 |
+
"usedImatrix": true,
|
| 48 |
+
"replacedArtifacts": [
|
| 49 |
+
{
|
| 50 |
+
"key": "118:0:0:0:0:0:0:0:0:0",
|
| 51 |
+
"shortName": "Unsloth-UD-Q6_K_XL",
|
| 52 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-UD-Q6_K_XL",
|
| 53 |
+
"kld": 0.001057,
|
| 54 |
+
"ppl": 5.611228,
|
| 55 |
+
"pplDeltaPercent": 0.06648238965670498,
|
| 56 |
+
"sizeBytes": 25963226816,
|
| 57 |
+
"sizeGB": 25.963226816,
|
| 58 |
+
"sizeGiB": 24.180139243602753,
|
| 59 |
+
"reasonCode": "STRICT_DOMINANCE",
|
| 60 |
+
"reason": "smart baseline-tuning strict dominance fallback: real benchmark validated lower KLD at same-or-smaller size"
|
| 61 |
+
}
|
| 62 |
+
]
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"key": "117:118:1:118:117:118:118:118:0:0",
|
| 66 |
+
"fileName": "Qwen3.8-27B-MQ-Q6_K_2.gguf",
|
| 67 |
+
"displayName": "MQ-Q6_K_2",
|
| 68 |
+
"shortName": "MQ-Q6_K_2",
|
| 69 |
+
"provider": "MagicQuant",
|
| 70 |
+
"quantFamily": "Q6_K",
|
| 71 |
+
"isHybrid": true,
|
| 72 |
+
"isExternalPureBaseline": false,
|
| 73 |
+
"isExternalRebuiltBaseline": false,
|
| 74 |
+
"isMaterializedTensorMapped": true,
|
| 75 |
+
"isExternalReference": false,
|
| 76 |
+
"downloadTarget": "./../../resolve/main/Qwen3.8-27B-MQ-Q6_K_2.gguf?download=true",
|
| 77 |
+
"kld": 0.001518,
|
| 78 |
+
"ppl": 5.613591,
|
| 79 |
+
"pplDeltaPercent": 0.10862238074008938,
|
| 80 |
+
"sizeBytes": 23213909696,
|
| 81 |
+
"sizeGB": 23.213909695999998,
|
| 82 |
+
"sizeGiB": 21.61963814496994,
|
| 83 |
+
"expectedSizeBytes": 23213909696,
|
| 84 |
+
"actualSizeBytes": 23213909696,
|
| 85 |
+
"usedImatrix": true,
|
| 86 |
+
"replacedArtifacts": [
|
| 87 |
+
{
|
| 88 |
+
"key": "117:0:0:0:0:0:0:0:0:0",
|
| 89 |
+
"shortName": "Unsloth-Q6_K",
|
| 90 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-Q6_K",
|
| 91 |
+
"kld": 0.001835,
|
| 92 |
+
"ppl": 5.608243,
|
| 93 |
+
"pplDeltaPercent": 0.013250111457867824,
|
| 94 |
+
"sizeBytes": 22884406976,
|
| 95 |
+
"sizeGB": 22.884406975999998,
|
| 96 |
+
"sizeGiB": 21.312764823436737,
|
| 97 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 98 |
+
"reason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium"
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"key": "117:118:118:118:117:118:118:118:0:0",
|
| 102 |
+
"shortName": "MQ-Unsloth-Q6_K",
|
| 103 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-Q6_K-D-Unsloth-Q6_K-E-Unsloth-Q6_K-H-Unsloth-Q6_K-K-Unsloth-UD-Q5_K_XL-O-Unsloth-Q6_K-Q-Unsloth-Q6_K-U-Unsloth-Q6_K",
|
| 104 |
+
"kld": 0.001723,
|
| 105 |
+
"ppl": 5.609167,
|
| 106 |
+
"pplDeltaPercent": 0.029728042799827145,
|
| 107 |
+
"sizeBytes": 22905992896,
|
| 108 |
+
"sizeGB": 22.905992896000004,
|
| 109 |
+
"sizeGiB": 21.33286827802658,
|
| 110 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 111 |
+
"reason": "meaningful spacing collapse; size gap below 599,772,979 bytes"
|
| 112 |
+
}
|
| 113 |
+
]
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"key": "116:117:117:117:117:1:117:117:0:0",
|
| 117 |
+
"fileName": "Qwen3.8-27B-MQ-Q5_K_1.gguf",
|
| 118 |
+
"displayName": "MQ-Q5_K_1",
|
| 119 |
+
"shortName": "MQ-Q5_K_1",
|
| 120 |
+
"provider": "MagicQuant",
|
| 121 |
+
"quantFamily": "Q5_K",
|
| 122 |
+
"isHybrid": true,
|
| 123 |
+
"isExternalPureBaseline": false,
|
| 124 |
+
"isExternalRebuiltBaseline": false,
|
| 125 |
+
"isMaterializedTensorMapped": true,
|
| 126 |
+
"isExternalReference": false,
|
| 127 |
+
"downloadTarget": "./../../resolve/main/Qwen3.8-27B-MQ-Q5_K_1.gguf?download=true",
|
| 128 |
+
"kld": 0.003955,
|
| 129 |
+
"ppl": 5.616773,
|
| 130 |
+
"pplDeltaPercent": 0.1653678109674598,
|
| 131 |
+
"sizeBytes": 20386768576,
|
| 132 |
+
"sizeGB": 20.386768576,
|
| 133 |
+
"sizeGiB": 18.986657798290253,
|
| 134 |
+
"expectedSizeBytes": 20386768576,
|
| 135 |
+
"actualSizeBytes": 20386768576,
|
| 136 |
+
"usedImatrix": true,
|
| 137 |
+
"replacedArtifacts": [
|
| 138 |
+
{
|
| 139 |
+
"key": "116:0:0:0:0:0:0:0:0:0",
|
| 140 |
+
"shortName": "Unsloth-UD-Q5_K_XL",
|
| 141 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-UD-Q5_K_XL",
|
| 142 |
+
"kld": 0.004133,
|
| 143 |
+
"ppl": 5.618565,
|
| 144 |
+
"pplDeltaPercent": 0.19732501114579268,
|
| 145 |
+
"sizeBytes": 20257253056,
|
| 146 |
+
"sizeGB": 20.257253056,
|
| 147 |
+
"sizeGiB": 18.86603707075119,
|
| 148 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 149 |
+
"reason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium"
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"key": "115:118:116:116:118:118:116:116:0:0",
|
| 153 |
+
"shortName": "MQ-Unsloth-Q5_K_M",
|
| 154 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-Q5_K_M-D-Unsloth-Q5_K_M-E-Unsloth-Q6_K-H-Unsloth-Q5_K_M-K-Unsloth-Q6_K-O-Unsloth-Q6_K-Q-Unsloth-Q5_K_M-U-Unsloth-Q5_K_M",
|
| 155 |
+
"kld": 0.004366,
|
| 156 |
+
"ppl": 5.614868,
|
| 157 |
+
"pplDeltaPercent": 0.1313954525189565,
|
| 158 |
+
"sizeBytes": 20091344576,
|
| 159 |
+
"sizeGB": 20.091344576,
|
| 160 |
+
"sizeGiB": 18.71152275800705,
|
| 161 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 162 |
+
"reason": "meaningful spacing collapse; size gap below 599,772,979 bytes"
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"key": "115:0:0:0:0:0:0:0:0:0",
|
| 166 |
+
"shortName": "Unsloth-Q5_K_M",
|
| 167 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-Q5_K_M",
|
| 168 |
+
"kld": 0.004814,
|
| 169 |
+
"ppl": 5.619185,
|
| 170 |
+
"pplDeltaPercent": 0.2083816317432001,
|
| 171 |
+
"sizeBytes": 19834054336,
|
| 172 |
+
"sizeGB": 19.834054335999998,
|
| 173 |
+
"sizeGiB": 18.471902549266815,
|
| 174 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 175 |
+
"reason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium"
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"key": "114:115:115:115:117:115:115:115:0:0",
|
| 179 |
+
"shortName": "MQ-Unsloth-Q5_K_S",
|
| 180 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-Q5_K_S-D-Unsloth-Q5_K_S-E-Unsloth-Q5_K_S-H-Unsloth-Q5_K_S-K-Unsloth-UD-Q5_K_XL-O-Unsloth-Q5_K_S-Q-Unsloth-Q5_K_S-U-Unsloth-Q5_K_S",
|
| 181 |
+
"kld": 0.005255,
|
| 182 |
+
"ppl": 5.62576,
|
| 183 |
+
"pplDeltaPercent": 0.32563530985287065,
|
| 184 |
+
"sizeBytes": 19315295936,
|
| 185 |
+
"sizeGB": 19.315295936000002,
|
| 186 |
+
"sizeGiB": 17.98877114057541,
|
| 187 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 188 |
+
"reason": "meaningful spacing collapse; size gap below 599,772,979 bytes"
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"key": "114:0:0:0:0:0:0:0:0:0",
|
| 192 |
+
"shortName": "Unsloth-Q5_K_S",
|
| 193 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-Q5_K_S",
|
| 194 |
+
"kld": 0.005358,
|
| 195 |
+
"ppl": 5.624662,
|
| 196 |
+
"pplDeltaPercent": 0.3060543914400339,
|
| 197 |
+
"sizeBytes": 19270035136,
|
| 198 |
+
"sizeGB": 19.270035135999997,
|
| 199 |
+
"sizeGiB": 17.946618735790253,
|
| 200 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 201 |
+
"reason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium"
|
| 202 |
+
}
|
| 203 |
+
]
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"key": "113:114:114:117:114:114:114:114:0:0",
|
| 207 |
+
"fileName": "Qwen3.8-27B-MQ-Q4_K_M_1.gguf",
|
| 208 |
+
"displayName": "MQ-Q4_K_M_1",
|
| 209 |
+
"shortName": "MQ-Q4_K_M_1",
|
| 210 |
+
"provider": "MagicQuant",
|
| 211 |
+
"quantFamily": "Q4_K_M",
|
| 212 |
+
"isHybrid": true,
|
| 213 |
+
"isExternalPureBaseline": false,
|
| 214 |
+
"isExternalRebuiltBaseline": false,
|
| 215 |
+
"isMaterializedTensorMapped": true,
|
| 216 |
+
"isExternalReference": false,
|
| 217 |
+
"downloadTarget": "./../../resolve/main/Qwen3.8-27B-MQ-Q4_K_M_1.gguf?download=true",
|
| 218 |
+
"kld": 0.008508,
|
| 219 |
+
"ppl": 5.611953,
|
| 220 |
+
"pplDeltaPercent": 0.07941150245206988,
|
| 221 |
+
"sizeBytes": 18105992896,
|
| 222 |
+
"sizeGB": 18.105992896000004,
|
| 223 |
+
"sizeGiB": 16.862519919872284,
|
| 224 |
+
"expectedSizeBytes": 18105992896,
|
| 225 |
+
"actualSizeBytes": 18105992896,
|
| 226 |
+
"usedImatrix": true,
|
| 227 |
+
"replacedArtifacts": [
|
| 228 |
+
{
|
| 229 |
+
"key": "113:0:0:0:0:0:0:0:0:0",
|
| 230 |
+
"shortName": "Unsloth-UD-Q4_K_XL",
|
| 231 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-UD-Q4_K_XL",
|
| 232 |
+
"kld": 0.009094,
|
| 233 |
+
"ppl": 5.610511,
|
| 234 |
+
"pplDeltaPercent": 0.05369594293356888,
|
| 235 |
+
"sizeBytes": 17963943616,
|
| 236 |
+
"sizeGB": 17.963943616,
|
| 237 |
+
"sizeGiB": 16.73022621870041,
|
| 238 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 239 |
+
"reason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium"
|
| 240 |
+
}
|
| 241 |
+
]
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"key": "112:113:113:110:113:113:113:115:0:0",
|
| 245 |
+
"fileName": "Qwen3.8-27B-MQ-IQ4_NL_1.gguf",
|
| 246 |
+
"displayName": "MQ-IQ4_NL_1",
|
| 247 |
+
"shortName": "MQ-IQ4_NL_1",
|
| 248 |
+
"provider": "MagicQuant",
|
| 249 |
+
"quantFamily": "IQ4_NL",
|
| 250 |
+
"isHybrid": true,
|
| 251 |
+
"isExternalPureBaseline": false,
|
| 252 |
+
"isExternalRebuiltBaseline": false,
|
| 253 |
+
"isMaterializedTensorMapped": true,
|
| 254 |
+
"isExternalReference": false,
|
| 255 |
+
"downloadTarget": "./../../resolve/main/Qwen3.8-27B-MQ-IQ4_NL_1.gguf?download=true",
|
| 256 |
+
"kld": 0.012627,
|
| 257 |
+
"ppl": 5.607022,
|
| 258 |
+
"pplDeltaPercent": -0.008524297815429346,
|
| 259 |
+
"sizeBytes": 17062823616,
|
| 260 |
+
"sizeGB": 17.062823616000003,
|
| 261 |
+
"sizeGiB": 15.890992820262909,
|
| 262 |
+
"expectedSizeBytes": 17062823616,
|
| 263 |
+
"actualSizeBytes": 17062823616,
|
| 264 |
+
"usedImatrix": true,
|
| 265 |
+
"replacedArtifacts": [
|
| 266 |
+
{
|
| 267 |
+
"key": "112:0:0:0:0:0:0:0:0:0",
|
| 268 |
+
"shortName": "Unsloth-Q4_K_M",
|
| 269 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-Q4_K_M",
|
| 270 |
+
"kld": 0.012959,
|
| 271 |
+
"ppl": 5.607668,
|
| 272 |
+
"pplDeltaPercent": 0.0029959875167256365,
|
| 273 |
+
"sizeBytes": 17106773696,
|
| 274 |
+
"sizeGB": 17.106773696,
|
| 275 |
+
"sizeGiB": 15.931924521923065,
|
| 276 |
+
"reasonCode": "STRICT_DOMINANCE",
|
| 277 |
+
"reason": "smart baseline-tuning strict dominance fallback: real benchmark validated lower KLD at same-or-smaller size"
|
| 278 |
+
}
|
| 279 |
+
]
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"key": "109:110:110:110:110:110:113:110:0:0",
|
| 283 |
+
"fileName": "Qwen3.8-27B-MQ-IQ4_XS_1.gguf",
|
| 284 |
+
"displayName": "MQ-IQ4_XS_1",
|
| 285 |
+
"shortName": "MQ-IQ4_XS_1",
|
| 286 |
+
"provider": "MagicQuant",
|
| 287 |
+
"quantFamily": "IQ4_XS",
|
| 288 |
+
"isHybrid": true,
|
| 289 |
+
"isExternalPureBaseline": false,
|
| 290 |
+
"isExternalRebuiltBaseline": false,
|
| 291 |
+
"isMaterializedTensorMapped": true,
|
| 292 |
+
"isExternalReference": false,
|
| 293 |
+
"downloadTarget": "./../../resolve/main/Qwen3.8-27B-MQ-IQ4_XS_1.gguf?download=true",
|
| 294 |
+
"kld": 0.013794,
|
| 295 |
+
"ppl": 5.641085,
|
| 296 |
+
"pplDeltaPercent": 0.5989300044583223,
|
| 297 |
+
"sizeBytes": 16337626816,
|
| 298 |
+
"sizeGB": 16.337626816,
|
| 299 |
+
"sizeGiB": 15.215600669384003,
|
| 300 |
+
"expectedSizeBytes": 16337626816,
|
| 301 |
+
"actualSizeBytes": 16337626816,
|
| 302 |
+
"usedImatrix": true,
|
| 303 |
+
"replacedArtifacts": [
|
| 304 |
+
{
|
| 305 |
+
"key": "109:0:0:0:0:0:0:0:0:0",
|
| 306 |
+
"shortName": "Unsloth-IQ4_NL",
|
| 307 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-IQ4_NL",
|
| 308 |
+
"kld": 0.014992,
|
| 309 |
+
"ppl": 5.654784,
|
| 310 |
+
"pplDeltaPercent": 0.8432278198840897,
|
| 311 |
+
"sizeBytes": 16337626816,
|
| 312 |
+
"sizeGB": 16.337626816,
|
| 313 |
+
"sizeGiB": 15.215600669384003,
|
| 314 |
+
"reasonCode": "STRICT_DOMINANCE",
|
| 315 |
+
"reason": "smart baseline-tuning strict dominance fallback: real benchmark validated lower KLD at same-or-smaller size"
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"key": "107:108:108:110:115:118:108:108:0:0",
|
| 319 |
+
"shortName": "MQ-Unsloth-IQ4_XS",
|
| 320 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-IQ4_XS-D-Unsloth-IQ4_XS-E-Unsloth-IQ4_XS-H-Unsloth-IQ4_XS-K-Unsloth-Q5_K_S-O-Unsloth-Q6_K-Q-Unsloth-IQ4_NL-U-Unsloth-IQ4_XS",
|
| 321 |
+
"kld": 0.014337,
|
| 322 |
+
"ppl": 5.650981,
|
| 323 |
+
"pplDeltaPercent": 0.7754079358002655,
|
| 324 |
+
"sizeBytes": 15904679616,
|
| 325 |
+
"sizeGB": 15.904679616000001,
|
| 326 |
+
"sizeGiB": 14.81238716840744,
|
| 327 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 328 |
+
"reason": "meaningful spacing collapse; size gap below 599,772,979 bytes"
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"key": "107:0:0:0:0:0:0:0:0:0",
|
| 332 |
+
"shortName": "Unsloth-IQ4_XS",
|
| 333 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-IQ4_XS",
|
| 334 |
+
"kld": 0.015238,
|
| 335 |
+
"ppl": 5.666636,
|
| 336 |
+
"pplDeltaPercent": 1.054587605884969,
|
| 337 |
+
"sizeBytes": 15705859776,
|
| 338 |
+
"sizeGB": 15.705859776,
|
| 339 |
+
"sizeGiB": 14.627221763134003,
|
| 340 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 341 |
+
"reason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium"
|
| 342 |
+
}
|
| 343 |
+
]
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"key": "106:0:0:0:0:0:0:0:0:0",
|
| 347 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-UD-Q3_K_XL.gguf",
|
| 348 |
+
"displayName": "UD-Unsloth-UD-Q3_K_XL",
|
| 349 |
+
"shortName": "UD-Unsloth-UD-Q3_K_XL",
|
| 350 |
+
"provider": "Unsloth",
|
| 351 |
+
"quantFamily": "UD-Unsloth-UD-Q3_K_XL",
|
| 352 |
+
"isHybrid": false,
|
| 353 |
+
"isExternalPureBaseline": true,
|
| 354 |
+
"isExternalRebuiltBaseline": false,
|
| 355 |
+
"isMaterializedTensorMapped": false,
|
| 356 |
+
"isExternalReference": false,
|
| 357 |
+
"downloadTarget": "./../../resolve/main/Qwen3.8-27B-UD-Unsloth-UD-Q3_K_XL.gguf?download=true",
|
| 358 |
+
"kld": 0.034902,
|
| 359 |
+
"ppl": 5.759128,
|
| 360 |
+
"pplDeltaPercent": 2.7040213999108276,
|
| 361 |
+
"sizeBytes": 13481977536,
|
| 362 |
+
"sizeGB": 13.481977536,
|
| 363 |
+
"sizeGiB": 12.556070029735565,
|
| 364 |
+
"expectedSizeBytes": 13481977536,
|
| 365 |
+
"actualSizeBytes": 13481977536,
|
| 366 |
+
"usedImatrix": true,
|
| 367 |
+
"replacedArtifacts": []
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"key": "103:104:104:104:109:104:104:104:0:0",
|
| 371 |
+
"fileName": "Qwen3.8-27B-MQ-IQ2_M_1.gguf",
|
| 372 |
+
"displayName": "MQ-IQ2_M_1",
|
| 373 |
+
"shortName": "MQ-IQ2_M_1",
|
| 374 |
+
"provider": "MagicQuant",
|
| 375 |
+
"quantFamily": "IQ2_M",
|
| 376 |
+
"isHybrid": true,
|
| 377 |
+
"isExternalPureBaseline": false,
|
| 378 |
+
"isExternalRebuiltBaseline": false,
|
| 379 |
+
"isMaterializedTensorMapped": true,
|
| 380 |
+
"isExternalReference": false,
|
| 381 |
+
"downloadTarget": "./../../resolve/main/Qwen3.8-27B-MQ-IQ2_M_1.gguf?download=true",
|
| 382 |
+
"kld": 0.057811,
|
| 383 |
+
"ppl": 5.8607,
|
| 384 |
+
"pplDeltaPercent": 4.5153811859117186,
|
| 385 |
+
"sizeBytes": 11960702656,
|
| 386 |
+
"sizeGB": 11.960702655999999,
|
| 387 |
+
"sizeGiB": 11.139272391796112,
|
| 388 |
+
"expectedSizeBytes": 11960702656,
|
| 389 |
+
"actualSizeBytes": 11960702656,
|
| 390 |
+
"usedImatrix": true,
|
| 391 |
+
"replacedArtifacts": [
|
| 392 |
+
{
|
| 393 |
+
"key": "103:0:0:0:0:0:0:0:0:0",
|
| 394 |
+
"shortName": "Unsloth-UD-IQ3_XXS",
|
| 395 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-UD-IQ3_XXS",
|
| 396 |
+
"kld": 0.058766,
|
| 397 |
+
"ppl": 5.876035,
|
| 398 |
+
"pplDeltaPercent": 4.788854213107445,
|
| 399 |
+
"sizeBytes": 11954476736,
|
| 400 |
+
"sizeGB": 11.954476735999998,
|
| 401 |
+
"sizeGiB": 11.133474051952362,
|
| 402 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 403 |
+
"reason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium"
|
| 404 |
+
}
|
| 405 |
+
]
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"key": "102:103:103:103:114:103:103:103:0:0",
|
| 409 |
+
"fileName": "Qwen3.8-27B-MQ-IQ2_M_2.gguf",
|
| 410 |
+
"displayName": "MQ-IQ2_M_2",
|
| 411 |
+
"shortName": "MQ-IQ2_M_2",
|
| 412 |
+
"provider": "MagicQuant",
|
| 413 |
+
"quantFamily": "IQ2_M",
|
| 414 |
+
"isHybrid": true,
|
| 415 |
+
"isExternalPureBaseline": false,
|
| 416 |
+
"isExternalRebuiltBaseline": false,
|
| 417 |
+
"isMaterializedTensorMapped": true,
|
| 418 |
+
"isExternalReference": false,
|
| 419 |
+
"downloadTarget": "./../../resolve/main/Qwen3.8-27B-MQ-IQ2_M_2.gguf?download=true",
|
| 420 |
+
"kld": 0.092394,
|
| 421 |
+
"ppl": 6.009092,
|
| 422 |
+
"pplDeltaPercent": 7.161694159607668,
|
| 423 |
+
"sizeBytes": 10769770176,
|
| 424 |
+
"sizeGB": 10.769770176000002,
|
| 425 |
+
"sizeGiB": 10.030130088329315,
|
| 426 |
+
"expectedSizeBytes": 10769770176,
|
| 427 |
+
"actualSizeBytes": 10769770176,
|
| 428 |
+
"usedImatrix": true,
|
| 429 |
+
"replacedArtifacts": [
|
| 430 |
+
{
|
| 431 |
+
"key": "102:0:0:0:0:0:0:0:0:0",
|
| 432 |
+
"shortName": "Unsloth-UD-Q2_K_XL",
|
| 433 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-UD-Q2_K_XL",
|
| 434 |
+
"kld": 0.094036,
|
| 435 |
+
"ppl": 6.019528,
|
| 436 |
+
"pplDeltaPercent": 7.347802050824794,
|
| 437 |
+
"sizeBytes": 10721027776,
|
| 438 |
+
"sizeGB": 10.721027776000001,
|
| 439 |
+
"sizeGiB": 9.984735190868378,
|
| 440 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 441 |
+
"reason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium"
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"key": "101:102:102:102:113:102:102:102:0:0",
|
| 445 |
+
"shortName": "MQ-Unsloth-UD-IQ2_M",
|
| 446 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-UD-IQ2_M-D-Unsloth-UD-IQ2_M-E-Unsloth-UD-IQ2_M-H-Unsloth-UD-IQ2_M-K-Unsloth-Q4_K_M-O-Unsloth-UD-IQ2_M-Q-Unsloth-UD-IQ2_M-U-Unsloth-UD-IQ2_M",
|
| 447 |
+
"kld": 0.108218,
|
| 448 |
+
"ppl": 6.097877,
|
| 449 |
+
"pplDeltaPercent": 8.745020062416415,
|
| 450 |
+
"sizeBytes": 10391299776,
|
| 451 |
+
"sizeGB": 10.391299776,
|
| 452 |
+
"sizeGiB": 9.677652060985565,
|
| 453 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 454 |
+
"reason": "meaningful spacing collapse; size gap below 599,772,979 bytes"
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"key": "101:0:0:0:0:0:0:0:0:0",
|
| 458 |
+
"shortName": "Unsloth-UD-IQ2_M",
|
| 459 |
+
"internalDisplayName": "Qwen3.8-27B-Qwen-Unsloth-UD-IQ2_M",
|
| 460 |
+
"kld": 0.109431,
|
| 461 |
+
"ppl": 6.109936,
|
| 462 |
+
"pplDeltaPercent": 8.960071333036119,
|
| 463 |
+
"sizeBytes": 10364511936,
|
| 464 |
+
"sizeGB": 10.364511936000001,
|
| 465 |
+
"sizeGiB": 9.652703940868378,
|
| 466 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 467 |
+
"reason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium"
|
| 468 |
+
}
|
| 469 |
+
]
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"key": "100:0:0:0:0:0:0:0:0:0",
|
| 473 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-UD-IQ2_XXS.gguf",
|
| 474 |
+
"displayName": "UD-Unsloth-UD-IQ2_XXS",
|
| 475 |
+
"shortName": "UD-Unsloth-UD-IQ2_XXS",
|
| 476 |
+
"provider": "Unsloth",
|
| 477 |
+
"quantFamily": "UD-Unsloth-UD-IQ2_XXS",
|
| 478 |
+
"isHybrid": false,
|
| 479 |
+
"isExternalPureBaseline": true,
|
| 480 |
+
"isExternalRebuiltBaseline": false,
|
| 481 |
+
"isMaterializedTensorMapped": false,
|
| 482 |
+
"isExternalReference": false,
|
| 483 |
+
"downloadTarget": "./../../resolve/main/Qwen3.8-27B-UD-Unsloth-UD-IQ2_XXS.gguf?download=true",
|
| 484 |
+
"kld": 0.182364,
|
| 485 |
+
"ppl": 6.480546,
|
| 486 |
+
"pplDeltaPercent": 15.569255461435585,
|
| 487 |
+
"sizeBytes": 9054652096,
|
| 488 |
+
"sizeGB": 9.054652096000002,
|
| 489 |
+
"sizeGiB": 8.432801902294159,
|
| 490 |
+
"expectedSizeBytes": 9054652096,
|
| 491 |
+
"actualSizeBytes": 9054652096,
|
| 492 |
+
"usedImatrix": true,
|
| 493 |
+
"replacedArtifacts": []
|
| 494 |
+
}
|
| 495 |
+
]
|
magicquant-manifest/magicquant.hybrid-map.json
ADDED
|
@@ -0,0 +1,370 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"ExportedFileName": "Qwen3.8-27B-MQ-Q6_K_1.gguf",
|
| 4 |
+
"DisplayName": "MQ-Q6_K_1",
|
| 5 |
+
"ProviderSource": "MagicQuant",
|
| 6 |
+
"BaselineFamily": "Q6_K",
|
| 7 |
+
"OriginalReferenceBaseline": "Unsloth-UD-Q6_K_XL",
|
| 8 |
+
"TensorGroups": {
|
| 9 |
+
"embeddings": "Unsloth-UD-Q6_K_XL",
|
| 10 |
+
"lm_head": "Unsloth-UD-Q6_K_XL",
|
| 11 |
+
"attn_q": "Unsloth-UD-Q6_K_XL",
|
| 12 |
+
"attn_kv": "Unsloth-UD-Q5_K_XL",
|
| 13 |
+
"attn_output": "Unsloth-UD-Q6_K_XL",
|
| 14 |
+
"ffn_up_gate": "Unsloth-UD-Q6_K_XL",
|
| 15 |
+
"ffn_down": "Unsloth-UD-Q6_K_XL"
|
| 16 |
+
},
|
| 17 |
+
"EffectiveQuantStateKey": "base=custom:qwen3.8-27b:unsloth/qwen3.8-27b-gguf:qwen3.8-27b-ud-q6_k_xl.gguf:232aedc0f6972d7a662cd49fdd947553355d0293421f529fc72332ba1203f94a|attn_kv=effective-map:c66920e650fba1bce6a7f2ecaa1394966a6e0395e9b18e18c533aa919ae863db|attn_output=effective:Q8_0|attn_q=effective:Q8_0|embeddings=effective:Q6_K|ffn_down=effective:Q8_0|ffn_up_gate=effective:Q6_K|lm_head=effective:Q8_0|moe_experts=base|moe_router=base",
|
| 18 |
+
"HasUnknownMappings": false,
|
| 19 |
+
"Warnings": [],
|
| 20 |
+
"UsedImatrix": true,
|
| 21 |
+
"ExpectedSizeBytes": 25941640896,
|
| 22 |
+
"ExpectedSizeGB": 25.941640896,
|
| 23 |
+
"ExpectedSizeGiB": 24.16003578901291,
|
| 24 |
+
"ActualSizeBytes": 25941640896,
|
| 25 |
+
"ActualSizeGB": 25.941640896,
|
| 26 |
+
"ActualSizeGiB": 24.16003578901291,
|
| 27 |
+
"OriginalExternalSource": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"ExportedFileName": "Qwen3.8-27B-MQ-Q6_K_2.gguf",
|
| 31 |
+
"DisplayName": "MQ-Q6_K_2",
|
| 32 |
+
"ProviderSource": "MagicQuant",
|
| 33 |
+
"BaselineFamily": "Q6_K",
|
| 34 |
+
"OriginalReferenceBaseline": "Unsloth-Q6_K",
|
| 35 |
+
"TensorGroups": {
|
| 36 |
+
"embeddings": "Unsloth-Q6_K",
|
| 37 |
+
"lm_head": "Q8_0",
|
| 38 |
+
"attn_q": "Unsloth-Q6_K",
|
| 39 |
+
"attn_kv": "Unsloth-UD-Q5_K_XL",
|
| 40 |
+
"attn_output": "Unsloth-Q6_K",
|
| 41 |
+
"ffn_up_gate": "Unsloth-Q6_K",
|
| 42 |
+
"ffn_down": "Unsloth-Q6_K"
|
| 43 |
+
},
|
| 44 |
+
"EffectiveQuantStateKey": "base=custom:qwen3.8-27b:unsloth/qwen3.8-27b-gguf:qwen3.8-27b-q6_k.gguf:34ed9335317b85d0819be2de3d7607d698f231df067f93f6002b6000004d7f00|attn_kv=effective-map:c66920e650fba1bce6a7f2ecaa1394966a6e0395e9b18e18c533aa919ae863db|attn_output=effective:Q6_K|attn_q=effective:Q6_K|embeddings=effective:Q6_K|ffn_down=effective:Q6_K|ffn_up_gate=effective:Q6_K|lm_head=effective:Q8_0|moe_experts=base|moe_router=base",
|
| 45 |
+
"HasUnknownMappings": false,
|
| 46 |
+
"Warnings": [],
|
| 47 |
+
"UsedImatrix": true,
|
| 48 |
+
"ExpectedSizeBytes": 23213909696,
|
| 49 |
+
"ExpectedSizeGB": 23.213909695999998,
|
| 50 |
+
"ExpectedSizeGiB": 21.61963814496994,
|
| 51 |
+
"ActualSizeBytes": 23213909696,
|
| 52 |
+
"ActualSizeGB": 23.213909695999998,
|
| 53 |
+
"ActualSizeGiB": 21.61963814496994,
|
| 54 |
+
"OriginalExternalSource": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"ExportedFileName": "Qwen3.8-27B-MQ-Q5_K_1.gguf",
|
| 58 |
+
"DisplayName": "MQ-Q5_K_1",
|
| 59 |
+
"ProviderSource": "MagicQuant",
|
| 60 |
+
"BaselineFamily": "Q5_K",
|
| 61 |
+
"OriginalReferenceBaseline": "Unsloth-UD-Q5_K_XL",
|
| 62 |
+
"TensorGroups": {
|
| 63 |
+
"embeddings": "Unsloth-UD-Q5_K_XL",
|
| 64 |
+
"lm_head": "Unsloth-UD-Q5_K_XL",
|
| 65 |
+
"attn_q": "Unsloth-UD-Q5_K_XL",
|
| 66 |
+
"attn_kv": "Unsloth-UD-Q5_K_XL",
|
| 67 |
+
"attn_output": "Q8_0",
|
| 68 |
+
"ffn_up_gate": "Unsloth-UD-Q5_K_XL",
|
| 69 |
+
"ffn_down": "Unsloth-UD-Q5_K_XL"
|
| 70 |
+
},
|
| 71 |
+
"EffectiveQuantStateKey": "base=custom:qwen3.8-27b:unsloth/qwen3.8-27b-gguf:qwen3.8-27b-ud-q5_k_xl.gguf:04dc0eff7b638f905e63fc1ca7a3255b412df5e996b049c0203744910b771c18|attn_kv=effective-map:c66920e650fba1bce6a7f2ecaa1394966a6e0395e9b18e18c533aa919ae863db|attn_output=effective:Q8_0|attn_q=effective-map:0053a621e96598bce4460d1012e339f388643c8d6903c7a4218db8edb6785dc3|embeddings=effective:Q5_K|ffn_down=effective:Q6_K|ffn_up_gate=effective:Q5_K|lm_head=effective:Q6_K|moe_experts=base|moe_router=base",
|
| 72 |
+
"HasUnknownMappings": false,
|
| 73 |
+
"Warnings": [],
|
| 74 |
+
"UsedImatrix": true,
|
| 75 |
+
"ExpectedSizeBytes": 20386768576,
|
| 76 |
+
"ExpectedSizeGB": 20.386768576,
|
| 77 |
+
"ExpectedSizeGiB": 18.986657798290253,
|
| 78 |
+
"ActualSizeBytes": 20386768576,
|
| 79 |
+
"ActualSizeGB": 20.386768576,
|
| 80 |
+
"ActualSizeGiB": 18.986657798290253,
|
| 81 |
+
"OriginalExternalSource": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"ExportedFileName": "Qwen3.8-27B-MQ-Q4_K_M_1.gguf",
|
| 85 |
+
"DisplayName": "MQ-Q4_K_M_1",
|
| 86 |
+
"ProviderSource": "MagicQuant",
|
| 87 |
+
"BaselineFamily": "Q4_K_M",
|
| 88 |
+
"OriginalReferenceBaseline": "Unsloth-UD-Q4_K_XL",
|
| 89 |
+
"TensorGroups": {
|
| 90 |
+
"embeddings": "Unsloth-UD-Q4_K_XL",
|
| 91 |
+
"lm_head": "Unsloth-UD-Q4_K_XL",
|
| 92 |
+
"attn_q": "Unsloth-UD-Q5_K_XL",
|
| 93 |
+
"attn_kv": "Unsloth-UD-Q4_K_XL",
|
| 94 |
+
"attn_output": "Unsloth-UD-Q4_K_XL",
|
| 95 |
+
"ffn_up_gate": "Unsloth-UD-Q4_K_XL",
|
| 96 |
+
"ffn_down": "Unsloth-UD-Q4_K_XL"
|
| 97 |
+
},
|
| 98 |
+
"EffectiveQuantStateKey": "base=custom:qwen3.8-27b:unsloth/qwen3.8-27b-gguf:qwen3.8-27b-ud-q4_k_xl.gguf:4cc12114fbc7bf0a8fa1ad97dcdae91b5923c438b9054dd26324b72a10084a47|attn_kv=effective-map:4c6434eb7f8875992cbcbcb6e5cb42adb963d7ab69b465a052d830d25b755735|attn_output=effective:Q5_K|attn_q=effective-map:0053a621e96598bce4460d1012e339f388643c8d6903c7a4218db8edb6785dc3|embeddings=effective:Q4_K|ffn_down=effective:Q5_K|ffn_up_gate=effective-map:6f2b36156b38d63d651e6bc64dc651d959e30c55c561cea09b12af215b39a2de|lm_head=effective:Q6_K|moe_experts=base|moe_router=base",
|
| 99 |
+
"HasUnknownMappings": false,
|
| 100 |
+
"Warnings": [],
|
| 101 |
+
"UsedImatrix": true,
|
| 102 |
+
"ExpectedSizeBytes": 18105992896,
|
| 103 |
+
"ExpectedSizeGB": 18.105992896000004,
|
| 104 |
+
"ExpectedSizeGiB": 16.862519919872284,
|
| 105 |
+
"ActualSizeBytes": 18105992896,
|
| 106 |
+
"ActualSizeGB": 18.105992896000004,
|
| 107 |
+
"ActualSizeGiB": 16.862519919872284,
|
| 108 |
+
"OriginalExternalSource": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"ExportedFileName": "Qwen3.8-27B-MQ-IQ4_NL_1.gguf",
|
| 112 |
+
"DisplayName": "MQ-IQ4_NL_1",
|
| 113 |
+
"ProviderSource": "MagicQuant",
|
| 114 |
+
"BaselineFamily": "IQ4_NL",
|
| 115 |
+
"OriginalReferenceBaseline": "Unsloth-Q4_K_M",
|
| 116 |
+
"TensorGroups": {
|
| 117 |
+
"embeddings": "Unsloth-Q4_K_M",
|
| 118 |
+
"lm_head": "Unsloth-Q4_K_M",
|
| 119 |
+
"attn_q": "Unsloth-IQ4_NL",
|
| 120 |
+
"attn_kv": "Unsloth-Q4_K_M",
|
| 121 |
+
"attn_output": "Unsloth-Q4_K_M",
|
| 122 |
+
"ffn_up_gate": "Unsloth-Q4_K_M",
|
| 123 |
+
"ffn_down": "Unsloth-Q5_K_S"
|
| 124 |
+
},
|
| 125 |
+
"EffectiveQuantStateKey": "base=custom:qwen3.8-27b:unsloth/qwen3.8-27b-gguf:qwen3.8-27b-q4_k_m.gguf:81ac0ff49a1400dc21a089afea19878cb0ef351c4a786a7d4781248628c45b55|attn_kv=effective-map:cfa3c7e22dab45e0a77aa3f407191b17d8ccab5105a1460e42258ae2afe0a200|attn_output=effective:Q4_K|attn_q=effective-map:e43136b54864568dfe78842f7bfcef25590cb8130b81a7dc7323cc805bf5054d|embeddings=effective:Q4_K|ffn_down=effective:Q5_K|ffn_up_gate=effective:Q4_K|lm_head=effective:Q6_K|moe_experts=base|moe_router=base",
|
| 126 |
+
"HasUnknownMappings": false,
|
| 127 |
+
"Warnings": [],
|
| 128 |
+
"UsedImatrix": true,
|
| 129 |
+
"ExpectedSizeBytes": 17062823616,
|
| 130 |
+
"ExpectedSizeGB": 17.062823616000003,
|
| 131 |
+
"ExpectedSizeGiB": 15.890992820262909,
|
| 132 |
+
"ActualSizeBytes": 17062823616,
|
| 133 |
+
"ActualSizeGB": 17.062823616000003,
|
| 134 |
+
"ActualSizeGiB": 15.890992820262909,
|
| 135 |
+
"OriginalExternalSource": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"ExportedFileName": "Qwen3.8-27B-MQ-IQ4_XS_1.gguf",
|
| 139 |
+
"DisplayName": "MQ-IQ4_XS_1",
|
| 140 |
+
"ProviderSource": "MagicQuant",
|
| 141 |
+
"BaselineFamily": "IQ4_XS",
|
| 142 |
+
"OriginalReferenceBaseline": "Unsloth-IQ4_NL",
|
| 143 |
+
"TensorGroups": {
|
| 144 |
+
"embeddings": "Unsloth-IQ4_NL",
|
| 145 |
+
"lm_head": "Unsloth-IQ4_NL",
|
| 146 |
+
"attn_q": "Unsloth-IQ4_NL",
|
| 147 |
+
"attn_kv": "Unsloth-IQ4_NL",
|
| 148 |
+
"attn_output": "Unsloth-IQ4_NL",
|
| 149 |
+
"ffn_up_gate": "Unsloth-Q4_K_M",
|
| 150 |
+
"ffn_down": "Unsloth-IQ4_NL"
|
| 151 |
+
},
|
| 152 |
+
"EffectiveQuantStateKey": "base=custom:qwen3.8-27b:unsloth/qwen3.8-27b-gguf:qwen3.8-27b-iq4_nl.gguf:00dfea690b294aed2a9e69b2f9d301e375710d2f30d02d333bbf0b5d101564d4|attn_kv=effective-map:2cd5cb986fcd2ae41531f3dc414cf9ab3ba5491f274ba58bcff791edb30616a9|attn_output=effective-map:383e223b3e3ef3921855bbbc0523636e06425f2f40bee5f3cdbdb5b2714518fa|attn_q=effective-map:e43136b54864568dfe78842f7bfcef25590cb8130b81a7dc7323cc805bf5054d|embeddings=effective:Q4_K|ffn_down=effective-map:24d6750b04df18a75532eeb901ab597a76400a9df46b08d1fbb21ff30bb4b1f3|ffn_up_gate=effective:Q4_K|lm_head=effective:Q6_K|moe_experts=base|moe_router=base",
|
| 153 |
+
"HasUnknownMappings": false,
|
| 154 |
+
"Warnings": [],
|
| 155 |
+
"UsedImatrix": true,
|
| 156 |
+
"ExpectedSizeBytes": 16337626816,
|
| 157 |
+
"ExpectedSizeGB": 16.337626816,
|
| 158 |
+
"ExpectedSizeGiB": 15.215600669384003,
|
| 159 |
+
"ActualSizeBytes": 16337626816,
|
| 160 |
+
"ActualSizeGB": 16.337626816,
|
| 161 |
+
"ActualSizeGiB": 15.215600669384003,
|
| 162 |
+
"OriginalExternalSource": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"ExportedFileName": "Qwen3.8-27B-MQ-IQ2_M_1.gguf",
|
| 166 |
+
"DisplayName": "MQ-IQ2_M_1",
|
| 167 |
+
"ProviderSource": "MagicQuant",
|
| 168 |
+
"BaselineFamily": "IQ2_M",
|
| 169 |
+
"OriginalReferenceBaseline": "Unsloth-UD-IQ3_XXS",
|
| 170 |
+
"TensorGroups": {
|
| 171 |
+
"embeddings": "Unsloth-UD-IQ3_XXS",
|
| 172 |
+
"lm_head": "Unsloth-UD-IQ3_XXS",
|
| 173 |
+
"attn_q": "Unsloth-UD-IQ3_XXS",
|
| 174 |
+
"attn_kv": "Unsloth-Q4_K_S",
|
| 175 |
+
"attn_output": "Unsloth-UD-IQ3_XXS",
|
| 176 |
+
"ffn_up_gate": "Unsloth-UD-IQ3_XXS",
|
| 177 |
+
"ffn_down": "Unsloth-UD-IQ3_XXS"
|
| 178 |
+
},
|
| 179 |
+
"EffectiveQuantStateKey": "base=custom:qwen3.8-27b:unsloth/qwen3.8-27b-gguf:qwen3.8-27b-ud-iq3_xxs.gguf:2cd579fcd506fe2b727b139668a821d1b4c0c8ebea771e17f1a4434e21f7203a|attn_kv=effective-map:9c255134fc941fed5bdfd21da67c95bdd0f027b1fd545533ca3ee82592875e0d|attn_output=effective:IQ3_S|attn_q=effective-map:4f52877ec1569763b619ba16f8b728f4fc0e15d0bc3fa5493e6d2a7dd66e6d13|embeddings=effective:unknown:Q2_K|ffn_down=effective:IQ3_S|ffn_up_gate=effective-map:ea33d3de7804e178645189aa31fe006d26af48d8fe9b269994d4d8f3be269e41|lm_head=effective:Q5_K|moe_experts=base|moe_router=base",
|
| 180 |
+
"HasUnknownMappings": true,
|
| 181 |
+
"Warnings": [
|
| 182 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 183 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 184 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 185 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 186 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 187 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 188 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 189 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 190 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 191 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 192 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 193 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 194 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 195 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 196 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 197 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 198 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 199 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 200 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 201 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 202 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 203 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 204 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 205 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 206 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 207 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 208 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 209 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 210 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 211 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 212 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 213 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 214 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 215 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 216 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 217 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 218 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 219 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 220 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 221 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 222 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 223 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 224 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 225 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 226 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 227 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 228 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 229 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 230 |
+
"Unknown or partially unmapped learned tensor state \u0027Q2_K\u0027 was preserved as raw metadata.",
|
| 231 |
+
"Unknown or partially unmapped learned tensor state \u0027Q2_K\u0027 was preserved as raw metadata."
|
| 232 |
+
],
|
| 233 |
+
"UsedImatrix": true,
|
| 234 |
+
"ExpectedSizeBytes": 11960702656,
|
| 235 |
+
"ExpectedSizeGB": 11.960702655999999,
|
| 236 |
+
"ExpectedSizeGiB": 11.139272391796112,
|
| 237 |
+
"ActualSizeBytes": 11960702656,
|
| 238 |
+
"ActualSizeGB": 11.960702655999999,
|
| 239 |
+
"ActualSizeGiB": 11.139272391796112,
|
| 240 |
+
"OriginalExternalSource": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"ExportedFileName": "Qwen3.8-27B-MQ-IQ2_M_2.gguf",
|
| 244 |
+
"DisplayName": "MQ-IQ2_M_2",
|
| 245 |
+
"ProviderSource": "MagicQuant",
|
| 246 |
+
"BaselineFamily": "IQ2_M",
|
| 247 |
+
"OriginalReferenceBaseline": "Unsloth-UD-Q2_K_XL",
|
| 248 |
+
"TensorGroups": {
|
| 249 |
+
"embeddings": "Unsloth-UD-Q2_K_XL",
|
| 250 |
+
"lm_head": "Unsloth-UD-Q2_K_XL",
|
| 251 |
+
"attn_q": "Unsloth-UD-Q2_K_XL",
|
| 252 |
+
"attn_kv": "Unsloth-UD-Q4_K_XL",
|
| 253 |
+
"attn_output": "Unsloth-UD-Q2_K_XL",
|
| 254 |
+
"ffn_up_gate": "Unsloth-UD-Q2_K_XL",
|
| 255 |
+
"ffn_down": "Unsloth-UD-Q2_K_XL"
|
| 256 |
+
},
|
| 257 |
+
"EffectiveQuantStateKey": "base=custom:qwen3.8-27b:unsloth/qwen3.8-27b-gguf:qwen3.8-27b-ud-q2_k_xl.gguf:b5bebf7e541e47a28e290c5d6a6f8386c4cd5aff49c2e2f221a373bcc1018b6f|attn_kv=effective-map:4c6434eb7f8875992cbcbcb6e5cb42adb963d7ab69b465a052d830d25b755735|attn_output=effective:IQ3_S|attn_q=effective-map:fb26f6e49dfbbf3e40336ec9c28f288af9f8b135b38848a42b496576bd01067a|embeddings=effective:unknown:Q2_K|ffn_down=effective-map:d46f77aa733e601b350720a0ceb16b444564b5049c2b6e9d778ea910329c24ed|ffn_up_gate=effective-map:b92537268535282a0eb1ee3f4f83dcee763d8d0935ad7f7ea35f2bfd6e536ce7|lm_head=effective:unknown:Q3_K|moe_experts=base|moe_router=base",
|
| 258 |
+
"HasUnknownMappings": true,
|
| 259 |
+
"Warnings": [
|
| 260 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 261 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 262 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 263 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 264 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 265 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 266 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 267 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 268 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 269 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 270 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 271 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 272 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 273 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 274 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 275 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 276 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 277 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 278 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 279 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 280 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 281 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 282 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 283 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 284 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 285 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 286 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 287 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 288 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 289 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 290 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 291 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 292 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 293 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 294 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 295 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 296 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 297 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 298 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 299 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 300 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 301 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 302 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 303 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 304 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 305 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 306 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 307 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 308 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 309 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 310 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 311 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 312 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 313 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 314 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 315 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 316 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 317 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 318 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 319 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 320 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 321 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 322 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 323 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 324 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 325 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 326 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 327 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 328 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 329 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 330 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 331 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 332 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 333 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 334 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 335 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 336 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 337 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 338 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 339 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 340 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 341 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 342 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 343 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 344 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 345 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 346 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 347 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 348 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 349 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 350 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 351 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 352 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 353 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 354 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 355 |
+
"Unknown or partially unmapped learned tensor state \u0027IQ1_M\u0027 was preserved as raw metadata.",
|
| 356 |
+
"Unknown or partially unmapped learned tensor state \u0027Q3_K\u0027 was preserved as raw metadata.",
|
| 357 |
+
"Unknown or partially unmapped learned tensor state \u0027Q2_K\u0027 was preserved as raw metadata.",
|
| 358 |
+
"Unknown or partially unmapped learned tensor state \u0027Q2_K\u0027 was preserved as raw metadata.",
|
| 359 |
+
"Unknown or partially unmapped learned tensor state \u0027Q3_K\u0027 was preserved as raw metadata."
|
| 360 |
+
],
|
| 361 |
+
"UsedImatrix": true,
|
| 362 |
+
"ExpectedSizeBytes": 10769770176,
|
| 363 |
+
"ExpectedSizeGB": 10.769770176000002,
|
| 364 |
+
"ExpectedSizeGiB": 10.030130088329315,
|
| 365 |
+
"ActualSizeBytes": 10769770176,
|
| 366 |
+
"ActualSizeGB": 10.769770176000002,
|
| 367 |
+
"ActualSizeGiB": 10.030130088329315,
|
| 368 |
+
"OriginalExternalSource": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
|
| 369 |
+
}
|
| 370 |
+
]
|
magicquant-manifest/magicquant.isolation-samples.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
magicquant-manifest/magicquant.replacements.json
ADDED
|
@@ -0,0 +1,1302 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"reasonCode": "STRICT_DOMINANCE",
|
| 4 |
+
"reasonDescription": "The winner was no larger and had lower real KLD than the removed anchor.",
|
| 5 |
+
"rawReason": "smart baseline-tuning strict dominance fallback: real benchmark validated lower KLD at same-or-smaller size",
|
| 6 |
+
"removed": {
|
| 7 |
+
"key": "118:0:0:0:0:0:0:0:0:0",
|
| 8 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-UD-Q6_K_XL.gguf",
|
| 9 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-UD-Q6_K_XL",
|
| 10 |
+
"shortName": "Unsloth-UD-Q6_K_XL",
|
| 11 |
+
"provider": "Unsloth",
|
| 12 |
+
"quantFamily": "Unsloth-UD-Q6_K_XL",
|
| 13 |
+
"isHybrid": false,
|
| 14 |
+
"isExternalPureBaseline": true,
|
| 15 |
+
"isExternalRebuiltBaseline": false,
|
| 16 |
+
"isMaterializedTensorMapped": false,
|
| 17 |
+
"kld": 0.001057,
|
| 18 |
+
"ppl": 5.611228,
|
| 19 |
+
"pplDeltaPercent": 0.06648238965670498,
|
| 20 |
+
"sizeBytes": 25963226816,
|
| 21 |
+
"sizeGB": 25.963226816,
|
| 22 |
+
"sizeGiB": 24.180139243602753
|
| 23 |
+
},
|
| 24 |
+
"winner": {
|
| 25 |
+
"key": "118:119:119:119:117:119:119:119:0:0",
|
| 26 |
+
"fileName": "Qwen3.8-27B-MQ-Q6_K_1.gguf",
|
| 27 |
+
"displayName": "MQ-Q6_K_1",
|
| 28 |
+
"shortName": "MQ-Q6_K_1",
|
| 29 |
+
"provider": "MagicQuant",
|
| 30 |
+
"quantFamily": "Q6_K",
|
| 31 |
+
"isHybrid": true,
|
| 32 |
+
"isExternalPureBaseline": false,
|
| 33 |
+
"isExternalRebuiltBaseline": false,
|
| 34 |
+
"isMaterializedTensorMapped": true,
|
| 35 |
+
"kld": 0.001052,
|
| 36 |
+
"ppl": 5.611099,
|
| 37 |
+
"pplDeltaPercent": 0.06418189924209278,
|
| 38 |
+
"sizeBytes": 25941640896,
|
| 39 |
+
"sizeGB": 25.941640896,
|
| 40 |
+
"sizeGiB": 24.16003578901291
|
| 41 |
+
},
|
| 42 |
+
"deltas": {
|
| 43 |
+
"kld": 5.000000000000013E-06,
|
| 44 |
+
"sizeBytes": 21585920,
|
| 45 |
+
"sizeGB": 0.021585919999999998,
|
| 46 |
+
"sizeGiB": 0.02010345458984375,
|
| 47 |
+
"removedPplDeltaPercent": 0.06648238965670498,
|
| 48 |
+
"winnerPplDeltaPercent": 0.06418189924209278,
|
| 49 |
+
"pplDeltaPercentImprovement": 0.002300490414612208
|
| 50 |
+
}
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 54 |
+
"reasonDescription": "Two candidates were too close in practical output space; the stronger one was kept.",
|
| 55 |
+
"rawReason": "meaningful spacing collapse; size gap below 599,772,979 bytes",
|
| 56 |
+
"removed": {
|
| 57 |
+
"key": "117:118:118:118:117:118:118:118:0:0",
|
| 58 |
+
"fileName": "Qwen3.8-27B-MQ-Q6_K.gguf",
|
| 59 |
+
"displayName": "Qwen3.8-27B-MQ-Q6_K",
|
| 60 |
+
"shortName": "MQ-Unsloth-Q6_K",
|
| 61 |
+
"provider": "MagicQuant",
|
| 62 |
+
"quantFamily": "Unsloth-Q6_K",
|
| 63 |
+
"isHybrid": true,
|
| 64 |
+
"isExternalPureBaseline": false,
|
| 65 |
+
"isExternalRebuiltBaseline": false,
|
| 66 |
+
"isMaterializedTensorMapped": true,
|
| 67 |
+
"kld": 0.001723,
|
| 68 |
+
"ppl": 5.609167,
|
| 69 |
+
"pplDeltaPercent": 0.029728042799827145,
|
| 70 |
+
"sizeBytes": 22905992896,
|
| 71 |
+
"sizeGB": 22.905992896000004,
|
| 72 |
+
"sizeGiB": 21.33286827802658
|
| 73 |
+
},
|
| 74 |
+
"winner": {
|
| 75 |
+
"key": "117:118:1:118:117:118:118:118:0:0",
|
| 76 |
+
"fileName": "Qwen3.8-27B-MQ-Q6_K_2.gguf",
|
| 77 |
+
"displayName": "MQ-Q6_K_2",
|
| 78 |
+
"shortName": "MQ-Q6_K_2",
|
| 79 |
+
"provider": "MagicQuant",
|
| 80 |
+
"quantFamily": "Q6_K",
|
| 81 |
+
"isHybrid": true,
|
| 82 |
+
"isExternalPureBaseline": false,
|
| 83 |
+
"isExternalRebuiltBaseline": false,
|
| 84 |
+
"isMaterializedTensorMapped": true,
|
| 85 |
+
"kld": 0.001518,
|
| 86 |
+
"ppl": 5.613591,
|
| 87 |
+
"pplDeltaPercent": 0.10862238074008938,
|
| 88 |
+
"sizeBytes": 23213909696,
|
| 89 |
+
"sizeGB": 23.213909695999998,
|
| 90 |
+
"sizeGiB": 21.61963814496994
|
| 91 |
+
},
|
| 92 |
+
"deltas": {
|
| 93 |
+
"kld": 0.0002049999999999999,
|
| 94 |
+
"sizeBytes": -307916800,
|
| 95 |
+
"sizeGB": -0.3079168,
|
| 96 |
+
"sizeGiB": -0.2867698669433594,
|
| 97 |
+
"removedPplDeltaPercent": 0.029728042799827145,
|
| 98 |
+
"winnerPplDeltaPercent": 0.10862238074008938,
|
| 99 |
+
"pplDeltaPercentImprovement": -0.07889433794026224
|
| 100 |
+
}
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 104 |
+
"reasonDescription": "The winner used only the configured near-baseline size premium and beat the real linear KLD trade line.",
|
| 105 |
+
"rawReason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium",
|
| 106 |
+
"removed": {
|
| 107 |
+
"key": "117:0:0:0:0:0:0:0:0:0",
|
| 108 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-Q6_K.gguf",
|
| 109 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-Q6_K",
|
| 110 |
+
"shortName": "Unsloth-Q6_K",
|
| 111 |
+
"provider": "Unsloth",
|
| 112 |
+
"quantFamily": "Unsloth-Q6_K",
|
| 113 |
+
"isHybrid": false,
|
| 114 |
+
"isExternalPureBaseline": true,
|
| 115 |
+
"isExternalRebuiltBaseline": false,
|
| 116 |
+
"isMaterializedTensorMapped": false,
|
| 117 |
+
"kld": 0.001835,
|
| 118 |
+
"ppl": 5.608243,
|
| 119 |
+
"pplDeltaPercent": 0.013250111457867824,
|
| 120 |
+
"sizeBytes": 22884406976,
|
| 121 |
+
"sizeGB": 22.884406975999998,
|
| 122 |
+
"sizeGiB": 21.312764823436737
|
| 123 |
+
},
|
| 124 |
+
"winner": {
|
| 125 |
+
"key": "117:118:1:118:117:118:118:118:0:0",
|
| 126 |
+
"fileName": "Qwen3.8-27B-MQ-Q6_K_2.gguf",
|
| 127 |
+
"displayName": "MQ-Q6_K_2",
|
| 128 |
+
"shortName": "MQ-Q6_K_2",
|
| 129 |
+
"provider": "MagicQuant",
|
| 130 |
+
"quantFamily": "Q6_K",
|
| 131 |
+
"isHybrid": true,
|
| 132 |
+
"isExternalPureBaseline": false,
|
| 133 |
+
"isExternalRebuiltBaseline": false,
|
| 134 |
+
"isMaterializedTensorMapped": true,
|
| 135 |
+
"kld": 0.001518,
|
| 136 |
+
"ppl": 5.613591,
|
| 137 |
+
"pplDeltaPercent": 0.10862238074008938,
|
| 138 |
+
"sizeBytes": 23213909696,
|
| 139 |
+
"sizeGB": 23.213909695999998,
|
| 140 |
+
"sizeGiB": 21.61963814496994
|
| 141 |
+
},
|
| 142 |
+
"deltas": {
|
| 143 |
+
"kld": 0.000317,
|
| 144 |
+
"sizeBytes": -329502720,
|
| 145 |
+
"sizeGB": -0.32950271999999997,
|
| 146 |
+
"sizeGiB": -0.3068733215332031,
|
| 147 |
+
"removedPplDeltaPercent": 0.013250111457867824,
|
| 148 |
+
"winnerPplDeltaPercent": 0.10862238074008938,
|
| 149 |
+
"pplDeltaPercentImprovement": -0.09537226928222156
|
| 150 |
+
}
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 154 |
+
"reasonDescription": "Two candidates were too close in practical output space; the stronger one was kept.",
|
| 155 |
+
"rawReason": "meaningful spacing collapse; size gap below 599,772,979 bytes",
|
| 156 |
+
"removed": {
|
| 157 |
+
"key": "117:0:0:0:0:0:0:0:0:0",
|
| 158 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-Q6_K.gguf",
|
| 159 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-Q6_K",
|
| 160 |
+
"shortName": "Unsloth-Q6_K",
|
| 161 |
+
"provider": "Unsloth",
|
| 162 |
+
"quantFamily": "Unsloth-Q6_K",
|
| 163 |
+
"isHybrid": false,
|
| 164 |
+
"isExternalPureBaseline": true,
|
| 165 |
+
"isExternalRebuiltBaseline": false,
|
| 166 |
+
"isMaterializedTensorMapped": false,
|
| 167 |
+
"kld": 0.001835,
|
| 168 |
+
"ppl": 5.608243,
|
| 169 |
+
"pplDeltaPercent": 0.013250111457867824,
|
| 170 |
+
"sizeBytes": 22884406976,
|
| 171 |
+
"sizeGB": 22.884406975999998,
|
| 172 |
+
"sizeGiB": 21.312764823436737
|
| 173 |
+
},
|
| 174 |
+
"winner": {
|
| 175 |
+
"key": "117:118:118:118:117:118:118:118:0:0",
|
| 176 |
+
"fileName": "Qwen3.8-27B-MQ-Q6_K.gguf",
|
| 177 |
+
"displayName": "Qwen3.8-27B-MQ-Q6_K",
|
| 178 |
+
"shortName": "MQ-Unsloth-Q6_K",
|
| 179 |
+
"provider": "MagicQuant",
|
| 180 |
+
"quantFamily": "Unsloth-Q6_K",
|
| 181 |
+
"isHybrid": true,
|
| 182 |
+
"isExternalPureBaseline": false,
|
| 183 |
+
"isExternalRebuiltBaseline": false,
|
| 184 |
+
"isMaterializedTensorMapped": true,
|
| 185 |
+
"kld": 0.001723,
|
| 186 |
+
"ppl": 5.609167,
|
| 187 |
+
"pplDeltaPercent": 0.029728042799827145,
|
| 188 |
+
"sizeBytes": 22905992896,
|
| 189 |
+
"sizeGB": 22.905992896000004,
|
| 190 |
+
"sizeGiB": 21.33286827802658
|
| 191 |
+
},
|
| 192 |
+
"deltas": {
|
| 193 |
+
"kld": 0.00011200000000000012,
|
| 194 |
+
"sizeBytes": -21585920,
|
| 195 |
+
"sizeGB": -0.021585919999999998,
|
| 196 |
+
"sizeGiB": -0.02010345458984375,
|
| 197 |
+
"removedPplDeltaPercent": 0.013250111457867824,
|
| 198 |
+
"winnerPplDeltaPercent": 0.029728042799827145,
|
| 199 |
+
"pplDeltaPercentImprovement": -0.016477931341959323
|
| 200 |
+
}
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 204 |
+
"reasonDescription": "The winner used only the configured near-baseline size premium and beat the real linear KLD trade line.",
|
| 205 |
+
"rawReason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium",
|
| 206 |
+
"removed": {
|
| 207 |
+
"key": "116:0:0:0:0:0:0:0:0:0",
|
| 208 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-UD-Q5_K_XL.gguf",
|
| 209 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-UD-Q5_K_XL",
|
| 210 |
+
"shortName": "Unsloth-UD-Q5_K_XL",
|
| 211 |
+
"provider": "Unsloth",
|
| 212 |
+
"quantFamily": "Unsloth-UD-Q5_K_XL",
|
| 213 |
+
"isHybrid": false,
|
| 214 |
+
"isExternalPureBaseline": true,
|
| 215 |
+
"isExternalRebuiltBaseline": false,
|
| 216 |
+
"isMaterializedTensorMapped": false,
|
| 217 |
+
"kld": 0.004133,
|
| 218 |
+
"ppl": 5.618565,
|
| 219 |
+
"pplDeltaPercent": 0.19732501114579268,
|
| 220 |
+
"sizeBytes": 20257253056,
|
| 221 |
+
"sizeGB": 20.257253056,
|
| 222 |
+
"sizeGiB": 18.86603707075119
|
| 223 |
+
},
|
| 224 |
+
"winner": {
|
| 225 |
+
"key": "116:117:117:117:117:1:117:117:0:0",
|
| 226 |
+
"fileName": "Qwen3.8-27B-MQ-Q5_K_1.gguf",
|
| 227 |
+
"displayName": "MQ-Q5_K_1",
|
| 228 |
+
"shortName": "MQ-Q5_K_1",
|
| 229 |
+
"provider": "MagicQuant",
|
| 230 |
+
"quantFamily": "Q5_K",
|
| 231 |
+
"isHybrid": true,
|
| 232 |
+
"isExternalPureBaseline": false,
|
| 233 |
+
"isExternalRebuiltBaseline": false,
|
| 234 |
+
"isMaterializedTensorMapped": true,
|
| 235 |
+
"kld": 0.003955,
|
| 236 |
+
"ppl": 5.616773,
|
| 237 |
+
"pplDeltaPercent": 0.1653678109674598,
|
| 238 |
+
"sizeBytes": 20386768576,
|
| 239 |
+
"sizeGB": 20.386768576,
|
| 240 |
+
"sizeGiB": 18.986657798290253
|
| 241 |
+
},
|
| 242 |
+
"deltas": {
|
| 243 |
+
"kld": 0.0001780000000000002,
|
| 244 |
+
"sizeBytes": -129515520,
|
| 245 |
+
"sizeGB": -0.12951552000000002,
|
| 246 |
+
"sizeGiB": -0.1206207275390625,
|
| 247 |
+
"removedPplDeltaPercent": 0.19732501114579268,
|
| 248 |
+
"winnerPplDeltaPercent": 0.1653678109674598,
|
| 249 |
+
"pplDeltaPercentImprovement": 0.03195720017833287
|
| 250 |
+
}
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 254 |
+
"reasonDescription": "Two candidates were too close in practical output space; the stronger one was kept.",
|
| 255 |
+
"rawReason": "meaningful spacing collapse; size gap below 599,772,979 bytes",
|
| 256 |
+
"removed": {
|
| 257 |
+
"key": "116:0:0:0:0:0:0:0:0:0",
|
| 258 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-UD-Q5_K_XL.gguf",
|
| 259 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-UD-Q5_K_XL",
|
| 260 |
+
"shortName": "Unsloth-UD-Q5_K_XL",
|
| 261 |
+
"provider": "Unsloth",
|
| 262 |
+
"quantFamily": "Unsloth-UD-Q5_K_XL",
|
| 263 |
+
"isHybrid": false,
|
| 264 |
+
"isExternalPureBaseline": true,
|
| 265 |
+
"isExternalRebuiltBaseline": false,
|
| 266 |
+
"isMaterializedTensorMapped": false,
|
| 267 |
+
"kld": 0.004133,
|
| 268 |
+
"ppl": 5.618565,
|
| 269 |
+
"pplDeltaPercent": 0.19732501114579268,
|
| 270 |
+
"sizeBytes": 20257253056,
|
| 271 |
+
"sizeGB": 20.257253056,
|
| 272 |
+
"sizeGiB": 18.86603707075119
|
| 273 |
+
},
|
| 274 |
+
"winner": {
|
| 275 |
+
"key": "116:117:117:117:117:1:117:117:0:0",
|
| 276 |
+
"fileName": "Qwen3.8-27B-MQ-Q5_K_1.gguf",
|
| 277 |
+
"displayName": "MQ-Q5_K_1",
|
| 278 |
+
"shortName": "MQ-Q5_K_1",
|
| 279 |
+
"provider": "MagicQuant",
|
| 280 |
+
"quantFamily": "Q5_K",
|
| 281 |
+
"isHybrid": true,
|
| 282 |
+
"isExternalPureBaseline": false,
|
| 283 |
+
"isExternalRebuiltBaseline": false,
|
| 284 |
+
"isMaterializedTensorMapped": true,
|
| 285 |
+
"kld": 0.003955,
|
| 286 |
+
"ppl": 5.616773,
|
| 287 |
+
"pplDeltaPercent": 0.1653678109674598,
|
| 288 |
+
"sizeBytes": 20386768576,
|
| 289 |
+
"sizeGB": 20.386768576,
|
| 290 |
+
"sizeGiB": 18.986657798290253
|
| 291 |
+
},
|
| 292 |
+
"deltas": {
|
| 293 |
+
"kld": 0.0001780000000000002,
|
| 294 |
+
"sizeBytes": -129515520,
|
| 295 |
+
"sizeGB": -0.12951552000000002,
|
| 296 |
+
"sizeGiB": -0.1206207275390625,
|
| 297 |
+
"removedPplDeltaPercent": 0.19732501114579268,
|
| 298 |
+
"winnerPplDeltaPercent": 0.1653678109674598,
|
| 299 |
+
"pplDeltaPercentImprovement": 0.03195720017833287
|
| 300 |
+
}
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 304 |
+
"reasonDescription": "Two candidates were too close in practical output space; the stronger one was kept.",
|
| 305 |
+
"rawReason": "meaningful spacing collapse; size gap below 599,772,979 bytes",
|
| 306 |
+
"removed": {
|
| 307 |
+
"key": "115:118:116:116:118:118:116:116:0:0",
|
| 308 |
+
"fileName": "Qwen3.8-27B-MQ-Q5_K.gguf",
|
| 309 |
+
"displayName": "Qwen3.8-27B-MQ-Q5_K",
|
| 310 |
+
"shortName": "MQ-Unsloth-Q5_K_M",
|
| 311 |
+
"provider": "MagicQuant",
|
| 312 |
+
"quantFamily": "Unsloth-Q5_K_M",
|
| 313 |
+
"isHybrid": true,
|
| 314 |
+
"isExternalPureBaseline": false,
|
| 315 |
+
"isExternalRebuiltBaseline": false,
|
| 316 |
+
"isMaterializedTensorMapped": true,
|
| 317 |
+
"kld": 0.004366,
|
| 318 |
+
"ppl": 5.614868,
|
| 319 |
+
"pplDeltaPercent": 0.1313954525189565,
|
| 320 |
+
"sizeBytes": 20091344576,
|
| 321 |
+
"sizeGB": 20.091344576,
|
| 322 |
+
"sizeGiB": 18.71152275800705
|
| 323 |
+
},
|
| 324 |
+
"winner": {
|
| 325 |
+
"key": "116:0:0:0:0:0:0:0:0:0",
|
| 326 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-UD-Q5_K_XL.gguf",
|
| 327 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-UD-Q5_K_XL",
|
| 328 |
+
"shortName": "Unsloth-UD-Q5_K_XL",
|
| 329 |
+
"provider": "Unsloth",
|
| 330 |
+
"quantFamily": "Unsloth-UD-Q5_K_XL",
|
| 331 |
+
"isHybrid": false,
|
| 332 |
+
"isExternalPureBaseline": true,
|
| 333 |
+
"isExternalRebuiltBaseline": false,
|
| 334 |
+
"isMaterializedTensorMapped": false,
|
| 335 |
+
"kld": 0.004133,
|
| 336 |
+
"ppl": 5.618565,
|
| 337 |
+
"pplDeltaPercent": 0.19732501114579268,
|
| 338 |
+
"sizeBytes": 20257253056,
|
| 339 |
+
"sizeGB": 20.257253056,
|
| 340 |
+
"sizeGiB": 18.86603707075119
|
| 341 |
+
},
|
| 342 |
+
"deltas": {
|
| 343 |
+
"kld": 0.0002329999999999997,
|
| 344 |
+
"sizeBytes": -165908480,
|
| 345 |
+
"sizeGB": -0.16590848,
|
| 346 |
+
"sizeGiB": -0.15451431274414062,
|
| 347 |
+
"removedPplDeltaPercent": 0.1313954525189565,
|
| 348 |
+
"winnerPplDeltaPercent": 0.19732501114579268,
|
| 349 |
+
"pplDeltaPercentImprovement": -0.06592955862683617
|
| 350 |
+
}
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 354 |
+
"reasonDescription": "The winner used only the configured near-baseline size premium and beat the real linear KLD trade line.",
|
| 355 |
+
"rawReason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium",
|
| 356 |
+
"removed": {
|
| 357 |
+
"key": "115:0:0:0:0:0:0:0:0:0",
|
| 358 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-Q5_K_M.gguf",
|
| 359 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-Q5_K_M",
|
| 360 |
+
"shortName": "Unsloth-Q5_K_M",
|
| 361 |
+
"provider": "Unsloth",
|
| 362 |
+
"quantFamily": "Unsloth-Q5_K_M",
|
| 363 |
+
"isHybrid": false,
|
| 364 |
+
"isExternalPureBaseline": true,
|
| 365 |
+
"isExternalRebuiltBaseline": false,
|
| 366 |
+
"isMaterializedTensorMapped": false,
|
| 367 |
+
"kld": 0.004814,
|
| 368 |
+
"ppl": 5.619185,
|
| 369 |
+
"pplDeltaPercent": 0.2083816317432001,
|
| 370 |
+
"sizeBytes": 19834054336,
|
| 371 |
+
"sizeGB": 19.834054335999998,
|
| 372 |
+
"sizeGiB": 18.471902549266815
|
| 373 |
+
},
|
| 374 |
+
"winner": {
|
| 375 |
+
"key": "115:118:116:116:118:118:116:116:0:0",
|
| 376 |
+
"fileName": "Qwen3.8-27B-MQ-Q5_K.gguf",
|
| 377 |
+
"displayName": "Qwen3.8-27B-MQ-Q5_K",
|
| 378 |
+
"shortName": "MQ-Unsloth-Q5_K_M",
|
| 379 |
+
"provider": "MagicQuant",
|
| 380 |
+
"quantFamily": "Unsloth-Q5_K_M",
|
| 381 |
+
"isHybrid": true,
|
| 382 |
+
"isExternalPureBaseline": false,
|
| 383 |
+
"isExternalRebuiltBaseline": false,
|
| 384 |
+
"isMaterializedTensorMapped": true,
|
| 385 |
+
"kld": 0.004366,
|
| 386 |
+
"ppl": 5.614868,
|
| 387 |
+
"pplDeltaPercent": 0.1313954525189565,
|
| 388 |
+
"sizeBytes": 20091344576,
|
| 389 |
+
"sizeGB": 20.091344576,
|
| 390 |
+
"sizeGiB": 18.71152275800705
|
| 391 |
+
},
|
| 392 |
+
"deltas": {
|
| 393 |
+
"kld": 0.0004479999999999996,
|
| 394 |
+
"sizeBytes": -257290240,
|
| 395 |
+
"sizeGB": -0.25729024,
|
| 396 |
+
"sizeGiB": -0.23962020874023438,
|
| 397 |
+
"removedPplDeltaPercent": 0.2083816317432001,
|
| 398 |
+
"winnerPplDeltaPercent": 0.1313954525189565,
|
| 399 |
+
"pplDeltaPercentImprovement": 0.07698617922424358
|
| 400 |
+
}
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 404 |
+
"reasonDescription": "Two candidates were too close in practical output space; the stronger one was kept.",
|
| 405 |
+
"rawReason": "meaningful spacing collapse; size gap below 599,772,979 bytes",
|
| 406 |
+
"removed": {
|
| 407 |
+
"key": "115:0:0:0:0:0:0:0:0:0",
|
| 408 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-Q5_K_M.gguf",
|
| 409 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-Q5_K_M",
|
| 410 |
+
"shortName": "Unsloth-Q5_K_M",
|
| 411 |
+
"provider": "Unsloth",
|
| 412 |
+
"quantFamily": "Unsloth-Q5_K_M",
|
| 413 |
+
"isHybrid": false,
|
| 414 |
+
"isExternalPureBaseline": true,
|
| 415 |
+
"isExternalRebuiltBaseline": false,
|
| 416 |
+
"isMaterializedTensorMapped": false,
|
| 417 |
+
"kld": 0.004814,
|
| 418 |
+
"ppl": 5.619185,
|
| 419 |
+
"pplDeltaPercent": 0.2083816317432001,
|
| 420 |
+
"sizeBytes": 19834054336,
|
| 421 |
+
"sizeGB": 19.834054335999998,
|
| 422 |
+
"sizeGiB": 18.471902549266815
|
| 423 |
+
},
|
| 424 |
+
"winner": {
|
| 425 |
+
"key": "115:118:116:116:118:118:116:116:0:0",
|
| 426 |
+
"fileName": "Qwen3.8-27B-MQ-Q5_K.gguf",
|
| 427 |
+
"displayName": "Qwen3.8-27B-MQ-Q5_K",
|
| 428 |
+
"shortName": "MQ-Unsloth-Q5_K_M",
|
| 429 |
+
"provider": "MagicQuant",
|
| 430 |
+
"quantFamily": "Unsloth-Q5_K_M",
|
| 431 |
+
"isHybrid": true,
|
| 432 |
+
"isExternalPureBaseline": false,
|
| 433 |
+
"isExternalRebuiltBaseline": false,
|
| 434 |
+
"isMaterializedTensorMapped": true,
|
| 435 |
+
"kld": 0.004366,
|
| 436 |
+
"ppl": 5.614868,
|
| 437 |
+
"pplDeltaPercent": 0.1313954525189565,
|
| 438 |
+
"sizeBytes": 20091344576,
|
| 439 |
+
"sizeGB": 20.091344576,
|
| 440 |
+
"sizeGiB": 18.71152275800705
|
| 441 |
+
},
|
| 442 |
+
"deltas": {
|
| 443 |
+
"kld": 0.0004479999999999996,
|
| 444 |
+
"sizeBytes": -257290240,
|
| 445 |
+
"sizeGB": -0.25729024,
|
| 446 |
+
"sizeGiB": -0.23962020874023438,
|
| 447 |
+
"removedPplDeltaPercent": 0.2083816317432001,
|
| 448 |
+
"winnerPplDeltaPercent": 0.1313954525189565,
|
| 449 |
+
"pplDeltaPercentImprovement": 0.07698617922424358
|
| 450 |
+
}
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 454 |
+
"reasonDescription": "Two candidates were too close in practical output space; the stronger one was kept.",
|
| 455 |
+
"rawReason": "meaningful spacing collapse; size gap below 599,772,979 bytes",
|
| 456 |
+
"removed": {
|
| 457 |
+
"key": "114:115:115:115:117:115:115:115:0:0",
|
| 458 |
+
"fileName": "Qwen3.8-27B-MQ-Q5_K_S.gguf",
|
| 459 |
+
"displayName": "Qwen3.8-27B-MQ-Q5_K_S",
|
| 460 |
+
"shortName": "MQ-Unsloth-Q5_K_S",
|
| 461 |
+
"provider": "MagicQuant",
|
| 462 |
+
"quantFamily": "Unsloth-Q5_K_S",
|
| 463 |
+
"isHybrid": true,
|
| 464 |
+
"isExternalPureBaseline": false,
|
| 465 |
+
"isExternalRebuiltBaseline": false,
|
| 466 |
+
"isMaterializedTensorMapped": true,
|
| 467 |
+
"kld": 0.005255,
|
| 468 |
+
"ppl": 5.62576,
|
| 469 |
+
"pplDeltaPercent": 0.32563530985287065,
|
| 470 |
+
"sizeBytes": 19315295936,
|
| 471 |
+
"sizeGB": 19.315295936000002,
|
| 472 |
+
"sizeGiB": 17.98877114057541
|
| 473 |
+
},
|
| 474 |
+
"winner": {
|
| 475 |
+
"key": "115:0:0:0:0:0:0:0:0:0",
|
| 476 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-Q5_K_M.gguf",
|
| 477 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-Q5_K_M",
|
| 478 |
+
"shortName": "Unsloth-Q5_K_M",
|
| 479 |
+
"provider": "Unsloth",
|
| 480 |
+
"quantFamily": "Unsloth-Q5_K_M",
|
| 481 |
+
"isHybrid": false,
|
| 482 |
+
"isExternalPureBaseline": true,
|
| 483 |
+
"isExternalRebuiltBaseline": false,
|
| 484 |
+
"isMaterializedTensorMapped": false,
|
| 485 |
+
"kld": 0.004814,
|
| 486 |
+
"ppl": 5.619185,
|
| 487 |
+
"pplDeltaPercent": 0.2083816317432001,
|
| 488 |
+
"sizeBytes": 19834054336,
|
| 489 |
+
"sizeGB": 19.834054335999998,
|
| 490 |
+
"sizeGiB": 18.471902549266815
|
| 491 |
+
},
|
| 492 |
+
"deltas": {
|
| 493 |
+
"kld": 0.0004410000000000004,
|
| 494 |
+
"sizeBytes": -518758400,
|
| 495 |
+
"sizeGB": -0.5187584000000001,
|
| 496 |
+
"sizeGiB": -0.48313140869140625,
|
| 497 |
+
"removedPplDeltaPercent": 0.32563530985287065,
|
| 498 |
+
"winnerPplDeltaPercent": 0.2083816317432001,
|
| 499 |
+
"pplDeltaPercentImprovement": 0.11725367810967055
|
| 500 |
+
}
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 504 |
+
"reasonDescription": "The winner used only the configured near-baseline size premium and beat the real linear KLD trade line.",
|
| 505 |
+
"rawReason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium",
|
| 506 |
+
"removed": {
|
| 507 |
+
"key": "114:0:0:0:0:0:0:0:0:0",
|
| 508 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-Q5_K_S.gguf",
|
| 509 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-Q5_K_S",
|
| 510 |
+
"shortName": "Unsloth-Q5_K_S",
|
| 511 |
+
"provider": "Unsloth",
|
| 512 |
+
"quantFamily": "Unsloth-Q5_K_S",
|
| 513 |
+
"isHybrid": false,
|
| 514 |
+
"isExternalPureBaseline": true,
|
| 515 |
+
"isExternalRebuiltBaseline": false,
|
| 516 |
+
"isMaterializedTensorMapped": false,
|
| 517 |
+
"kld": 0.005358,
|
| 518 |
+
"ppl": 5.624662,
|
| 519 |
+
"pplDeltaPercent": 0.3060543914400339,
|
| 520 |
+
"sizeBytes": 19270035136,
|
| 521 |
+
"sizeGB": 19.270035135999997,
|
| 522 |
+
"sizeGiB": 17.946618735790253
|
| 523 |
+
},
|
| 524 |
+
"winner": {
|
| 525 |
+
"key": "114:115:115:115:117:115:115:115:0:0",
|
| 526 |
+
"fileName": "Qwen3.8-27B-MQ-Q5_K_S.gguf",
|
| 527 |
+
"displayName": "Qwen3.8-27B-MQ-Q5_K_S",
|
| 528 |
+
"shortName": "MQ-Unsloth-Q5_K_S",
|
| 529 |
+
"provider": "MagicQuant",
|
| 530 |
+
"quantFamily": "Unsloth-Q5_K_S",
|
| 531 |
+
"isHybrid": true,
|
| 532 |
+
"isExternalPureBaseline": false,
|
| 533 |
+
"isExternalRebuiltBaseline": false,
|
| 534 |
+
"isMaterializedTensorMapped": true,
|
| 535 |
+
"kld": 0.005255,
|
| 536 |
+
"ppl": 5.62576,
|
| 537 |
+
"pplDeltaPercent": 0.32563530985287065,
|
| 538 |
+
"sizeBytes": 19315295936,
|
| 539 |
+
"sizeGB": 19.315295936000002,
|
| 540 |
+
"sizeGiB": 17.98877114057541
|
| 541 |
+
},
|
| 542 |
+
"deltas": {
|
| 543 |
+
"kld": 0.00010299999999999979,
|
| 544 |
+
"sizeBytes": -45260800,
|
| 545 |
+
"sizeGB": -0.045260800000000004,
|
| 546 |
+
"sizeGiB": -0.04215240478515625,
|
| 547 |
+
"removedPplDeltaPercent": 0.3060543914400339,
|
| 548 |
+
"winnerPplDeltaPercent": 0.32563530985287065,
|
| 549 |
+
"pplDeltaPercentImprovement": -0.019580918412836767
|
| 550 |
+
}
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 554 |
+
"reasonDescription": "Two candidates were too close in practical output space; the stronger one was kept.",
|
| 555 |
+
"rawReason": "meaningful spacing collapse; size gap below 599,772,979 bytes",
|
| 556 |
+
"removed": {
|
| 557 |
+
"key": "114:0:0:0:0:0:0:0:0:0",
|
| 558 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-Q5_K_S.gguf",
|
| 559 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-Q5_K_S",
|
| 560 |
+
"shortName": "Unsloth-Q5_K_S",
|
| 561 |
+
"provider": "Unsloth",
|
| 562 |
+
"quantFamily": "Unsloth-Q5_K_S",
|
| 563 |
+
"isHybrid": false,
|
| 564 |
+
"isExternalPureBaseline": true,
|
| 565 |
+
"isExternalRebuiltBaseline": false,
|
| 566 |
+
"isMaterializedTensorMapped": false,
|
| 567 |
+
"kld": 0.005358,
|
| 568 |
+
"ppl": 5.624662,
|
| 569 |
+
"pplDeltaPercent": 0.3060543914400339,
|
| 570 |
+
"sizeBytes": 19270035136,
|
| 571 |
+
"sizeGB": 19.270035135999997,
|
| 572 |
+
"sizeGiB": 17.946618735790253
|
| 573 |
+
},
|
| 574 |
+
"winner": {
|
| 575 |
+
"key": "114:115:115:115:117:115:115:115:0:0",
|
| 576 |
+
"fileName": "Qwen3.8-27B-MQ-Q5_K_S.gguf",
|
| 577 |
+
"displayName": "Qwen3.8-27B-MQ-Q5_K_S",
|
| 578 |
+
"shortName": "MQ-Unsloth-Q5_K_S",
|
| 579 |
+
"provider": "MagicQuant",
|
| 580 |
+
"quantFamily": "Unsloth-Q5_K_S",
|
| 581 |
+
"isHybrid": true,
|
| 582 |
+
"isExternalPureBaseline": false,
|
| 583 |
+
"isExternalRebuiltBaseline": false,
|
| 584 |
+
"isMaterializedTensorMapped": true,
|
| 585 |
+
"kld": 0.005255,
|
| 586 |
+
"ppl": 5.62576,
|
| 587 |
+
"pplDeltaPercent": 0.32563530985287065,
|
| 588 |
+
"sizeBytes": 19315295936,
|
| 589 |
+
"sizeGB": 19.315295936000002,
|
| 590 |
+
"sizeGiB": 17.98877114057541
|
| 591 |
+
},
|
| 592 |
+
"deltas": {
|
| 593 |
+
"kld": 0.00010299999999999979,
|
| 594 |
+
"sizeBytes": -45260800,
|
| 595 |
+
"sizeGB": -0.045260800000000004,
|
| 596 |
+
"sizeGiB": -0.04215240478515625,
|
| 597 |
+
"removedPplDeltaPercent": 0.3060543914400339,
|
| 598 |
+
"winnerPplDeltaPercent": 0.32563530985287065,
|
| 599 |
+
"pplDeltaPercentImprovement": -0.019580918412836767
|
| 600 |
+
}
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 604 |
+
"reasonDescription": "The winner used only the configured near-baseline size premium and beat the real linear KLD trade line.",
|
| 605 |
+
"rawReason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium",
|
| 606 |
+
"removed": {
|
| 607 |
+
"key": "113:0:0:0:0:0:0:0:0:0",
|
| 608 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-UD-Q4_K_XL.gguf",
|
| 609 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-UD-Q4_K_XL",
|
| 610 |
+
"shortName": "Unsloth-UD-Q4_K_XL",
|
| 611 |
+
"provider": "Unsloth",
|
| 612 |
+
"quantFamily": "Unsloth-UD-Q4_K_XL",
|
| 613 |
+
"isHybrid": false,
|
| 614 |
+
"isExternalPureBaseline": true,
|
| 615 |
+
"isExternalRebuiltBaseline": false,
|
| 616 |
+
"isMaterializedTensorMapped": false,
|
| 617 |
+
"kld": 0.009094,
|
| 618 |
+
"ppl": 5.610511,
|
| 619 |
+
"pplDeltaPercent": 0.05369594293356888,
|
| 620 |
+
"sizeBytes": 17963943616,
|
| 621 |
+
"sizeGB": 17.963943616,
|
| 622 |
+
"sizeGiB": 16.73022621870041
|
| 623 |
+
},
|
| 624 |
+
"winner": {
|
| 625 |
+
"key": "113:114:114:117:114:114:114:114:0:0",
|
| 626 |
+
"fileName": "Qwen3.8-27B-MQ-Q4_K_M_1.gguf",
|
| 627 |
+
"displayName": "MQ-Q4_K_M_1",
|
| 628 |
+
"shortName": "MQ-Q4_K_M_1",
|
| 629 |
+
"provider": "MagicQuant",
|
| 630 |
+
"quantFamily": "Q4_K_M",
|
| 631 |
+
"isHybrid": true,
|
| 632 |
+
"isExternalPureBaseline": false,
|
| 633 |
+
"isExternalRebuiltBaseline": false,
|
| 634 |
+
"isMaterializedTensorMapped": true,
|
| 635 |
+
"kld": 0.008508,
|
| 636 |
+
"ppl": 5.611953,
|
| 637 |
+
"pplDeltaPercent": 0.07941150245206988,
|
| 638 |
+
"sizeBytes": 18105992896,
|
| 639 |
+
"sizeGB": 18.105992896000004,
|
| 640 |
+
"sizeGiB": 16.862519919872284
|
| 641 |
+
},
|
| 642 |
+
"deltas": {
|
| 643 |
+
"kld": 0.0005859999999999997,
|
| 644 |
+
"sizeBytes": -142049280,
|
| 645 |
+
"sizeGB": -0.14204928,
|
| 646 |
+
"sizeGiB": -0.132293701171875,
|
| 647 |
+
"removedPplDeltaPercent": 0.05369594293356888,
|
| 648 |
+
"winnerPplDeltaPercent": 0.07941150245206988,
|
| 649 |
+
"pplDeltaPercentImprovement": -0.025715559518500998
|
| 650 |
+
}
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 654 |
+
"reasonDescription": "Two candidates were too close in practical output space; the stronger one was kept.",
|
| 655 |
+
"rawReason": "meaningful spacing collapse; size gap below 599,772,979 bytes",
|
| 656 |
+
"removed": {
|
| 657 |
+
"key": "113:0:0:0:0:0:0:0:0:0",
|
| 658 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-UD-Q4_K_XL.gguf",
|
| 659 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-UD-Q4_K_XL",
|
| 660 |
+
"shortName": "Unsloth-UD-Q4_K_XL",
|
| 661 |
+
"provider": "Unsloth",
|
| 662 |
+
"quantFamily": "Unsloth-UD-Q4_K_XL",
|
| 663 |
+
"isHybrid": false,
|
| 664 |
+
"isExternalPureBaseline": true,
|
| 665 |
+
"isExternalRebuiltBaseline": false,
|
| 666 |
+
"isMaterializedTensorMapped": false,
|
| 667 |
+
"kld": 0.009094,
|
| 668 |
+
"ppl": 5.610511,
|
| 669 |
+
"pplDeltaPercent": 0.05369594293356888,
|
| 670 |
+
"sizeBytes": 17963943616,
|
| 671 |
+
"sizeGB": 17.963943616,
|
| 672 |
+
"sizeGiB": 16.73022621870041
|
| 673 |
+
},
|
| 674 |
+
"winner": {
|
| 675 |
+
"key": "113:114:114:117:114:114:114:114:0:0",
|
| 676 |
+
"fileName": "Qwen3.8-27B-MQ-Q4_K_M_1.gguf",
|
| 677 |
+
"displayName": "MQ-Q4_K_M_1",
|
| 678 |
+
"shortName": "MQ-Q4_K_M_1",
|
| 679 |
+
"provider": "MagicQuant",
|
| 680 |
+
"quantFamily": "Q4_K_M",
|
| 681 |
+
"isHybrid": true,
|
| 682 |
+
"isExternalPureBaseline": false,
|
| 683 |
+
"isExternalRebuiltBaseline": false,
|
| 684 |
+
"isMaterializedTensorMapped": true,
|
| 685 |
+
"kld": 0.008508,
|
| 686 |
+
"ppl": 5.611953,
|
| 687 |
+
"pplDeltaPercent": 0.07941150245206988,
|
| 688 |
+
"sizeBytes": 18105992896,
|
| 689 |
+
"sizeGB": 18.105992896000004,
|
| 690 |
+
"sizeGiB": 16.862519919872284
|
| 691 |
+
},
|
| 692 |
+
"deltas": {
|
| 693 |
+
"kld": 0.0005859999999999997,
|
| 694 |
+
"sizeBytes": -142049280,
|
| 695 |
+
"sizeGB": -0.14204928,
|
| 696 |
+
"sizeGiB": -0.132293701171875,
|
| 697 |
+
"removedPplDeltaPercent": 0.05369594293356888,
|
| 698 |
+
"winnerPplDeltaPercent": 0.07941150245206988,
|
| 699 |
+
"pplDeltaPercentImprovement": -0.025715559518500998
|
| 700 |
+
}
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"reasonCode": "STRICT_DOMINANCE",
|
| 704 |
+
"reasonDescription": "The winner was no larger and had lower real KLD than the removed anchor.",
|
| 705 |
+
"rawReason": "smart baseline-tuning strict dominance fallback: real benchmark validated lower KLD at same-or-smaller size",
|
| 706 |
+
"removed": {
|
| 707 |
+
"key": "112:0:0:0:0:0:0:0:0:0",
|
| 708 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-Q4_K_M.gguf",
|
| 709 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-Q4_K_M",
|
| 710 |
+
"shortName": "Unsloth-Q4_K_M",
|
| 711 |
+
"provider": "Unsloth",
|
| 712 |
+
"quantFamily": "Unsloth-Q4_K_M",
|
| 713 |
+
"isHybrid": false,
|
| 714 |
+
"isExternalPureBaseline": true,
|
| 715 |
+
"isExternalRebuiltBaseline": false,
|
| 716 |
+
"isMaterializedTensorMapped": false,
|
| 717 |
+
"kld": 0.012959,
|
| 718 |
+
"ppl": 5.607668,
|
| 719 |
+
"pplDeltaPercent": 0.0029959875167256365,
|
| 720 |
+
"sizeBytes": 17106773696,
|
| 721 |
+
"sizeGB": 17.106773696,
|
| 722 |
+
"sizeGiB": 15.931924521923065
|
| 723 |
+
},
|
| 724 |
+
"winner": {
|
| 725 |
+
"key": "112:113:113:110:113:113:113:115:0:0",
|
| 726 |
+
"fileName": "Qwen3.8-27B-MQ-IQ4_NL_1.gguf",
|
| 727 |
+
"displayName": "MQ-IQ4_NL_1",
|
| 728 |
+
"shortName": "MQ-IQ4_NL_1",
|
| 729 |
+
"provider": "MagicQuant",
|
| 730 |
+
"quantFamily": "IQ4_NL",
|
| 731 |
+
"isHybrid": true,
|
| 732 |
+
"isExternalPureBaseline": false,
|
| 733 |
+
"isExternalRebuiltBaseline": false,
|
| 734 |
+
"isMaterializedTensorMapped": true,
|
| 735 |
+
"kld": 0.012627,
|
| 736 |
+
"ppl": 5.607022,
|
| 737 |
+
"pplDeltaPercent": -0.008524297815429346,
|
| 738 |
+
"sizeBytes": 17062823616,
|
| 739 |
+
"sizeGB": 17.062823616000003,
|
| 740 |
+
"sizeGiB": 15.890992820262909
|
| 741 |
+
},
|
| 742 |
+
"deltas": {
|
| 743 |
+
"kld": 0.0003320000000000007,
|
| 744 |
+
"sizeBytes": 43950080,
|
| 745 |
+
"sizeGB": 0.04395008,
|
| 746 |
+
"sizeGiB": 0.04093170166015625,
|
| 747 |
+
"removedPplDeltaPercent": 0.0029959875167256365,
|
| 748 |
+
"winnerPplDeltaPercent": -0.008524297815429346,
|
| 749 |
+
"pplDeltaPercentImprovement": 0.011520285332154982
|
| 750 |
+
}
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 754 |
+
"reasonDescription": "Two candidates were too close in practical output space; the stronger one was kept.",
|
| 755 |
+
"rawReason": "meaningful spacing collapse; size gap below 599,772,979 bytes",
|
| 756 |
+
"removed": {
|
| 757 |
+
"key": "107:108:108:110:115:118:108:108:0:0",
|
| 758 |
+
"fileName": "Qwen3.8-27B-MQ-IQ4_XS.gguf",
|
| 759 |
+
"displayName": "Qwen3.8-27B-MQ-IQ4_XS",
|
| 760 |
+
"shortName": "MQ-Unsloth-IQ4_XS",
|
| 761 |
+
"provider": "MagicQuant",
|
| 762 |
+
"quantFamily": "Unsloth-IQ4_XS",
|
| 763 |
+
"isHybrid": true,
|
| 764 |
+
"isExternalPureBaseline": false,
|
| 765 |
+
"isExternalRebuiltBaseline": false,
|
| 766 |
+
"isMaterializedTensorMapped": true,
|
| 767 |
+
"kld": 0.014337,
|
| 768 |
+
"ppl": 5.650981,
|
| 769 |
+
"pplDeltaPercent": 0.7754079358002655,
|
| 770 |
+
"sizeBytes": 15904679616,
|
| 771 |
+
"sizeGB": 15.904679616000001,
|
| 772 |
+
"sizeGiB": 14.81238716840744
|
| 773 |
+
},
|
| 774 |
+
"winner": {
|
| 775 |
+
"key": "109:110:110:110:110:110:113:110:0:0",
|
| 776 |
+
"fileName": "Qwen3.8-27B-MQ-IQ4_XS_1.gguf",
|
| 777 |
+
"displayName": "MQ-IQ4_XS_1",
|
| 778 |
+
"shortName": "MQ-IQ4_XS_1",
|
| 779 |
+
"provider": "MagicQuant",
|
| 780 |
+
"quantFamily": "IQ4_XS",
|
| 781 |
+
"isHybrid": true,
|
| 782 |
+
"isExternalPureBaseline": false,
|
| 783 |
+
"isExternalRebuiltBaseline": false,
|
| 784 |
+
"isMaterializedTensorMapped": true,
|
| 785 |
+
"kld": 0.013794,
|
| 786 |
+
"ppl": 5.641085,
|
| 787 |
+
"pplDeltaPercent": 0.5989300044583223,
|
| 788 |
+
"sizeBytes": 16337626816,
|
| 789 |
+
"sizeGB": 16.337626816,
|
| 790 |
+
"sizeGiB": 15.215600669384003
|
| 791 |
+
},
|
| 792 |
+
"deltas": {
|
| 793 |
+
"kld": 0.0005430000000000001,
|
| 794 |
+
"sizeBytes": -432947200,
|
| 795 |
+
"sizeGB": -0.43294720000000003,
|
| 796 |
+
"sizeGiB": -0.4032135009765625,
|
| 797 |
+
"removedPplDeltaPercent": 0.7754079358002655,
|
| 798 |
+
"winnerPplDeltaPercent": 0.5989300044583223,
|
| 799 |
+
"pplDeltaPercentImprovement": 0.17647793134194312
|
| 800 |
+
}
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"reasonCode": "STRICT_DOMINANCE",
|
| 804 |
+
"reasonDescription": "The winner was no larger and had lower real KLD than the removed anchor.",
|
| 805 |
+
"rawReason": "smart baseline-tuning strict dominance fallback: real benchmark validated lower KLD at same-or-smaller size",
|
| 806 |
+
"removed": {
|
| 807 |
+
"key": "109:0:0:0:0:0:0:0:0:0",
|
| 808 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-IQ4_NL.gguf",
|
| 809 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-IQ4_NL",
|
| 810 |
+
"shortName": "Unsloth-IQ4_NL",
|
| 811 |
+
"provider": "Unsloth",
|
| 812 |
+
"quantFamily": "Unsloth-IQ4_NL",
|
| 813 |
+
"isHybrid": false,
|
| 814 |
+
"isExternalPureBaseline": true,
|
| 815 |
+
"isExternalRebuiltBaseline": false,
|
| 816 |
+
"isMaterializedTensorMapped": false,
|
| 817 |
+
"kld": 0.014992,
|
| 818 |
+
"ppl": 5.654784,
|
| 819 |
+
"pplDeltaPercent": 0.8432278198840897,
|
| 820 |
+
"sizeBytes": 16337626816,
|
| 821 |
+
"sizeGB": 16.337626816,
|
| 822 |
+
"sizeGiB": 15.215600669384003
|
| 823 |
+
},
|
| 824 |
+
"winner": {
|
| 825 |
+
"key": "109:110:110:110:110:110:113:110:0:0",
|
| 826 |
+
"fileName": "Qwen3.8-27B-MQ-IQ4_XS_1.gguf",
|
| 827 |
+
"displayName": "MQ-IQ4_XS_1",
|
| 828 |
+
"shortName": "MQ-IQ4_XS_1",
|
| 829 |
+
"provider": "MagicQuant",
|
| 830 |
+
"quantFamily": "IQ4_XS",
|
| 831 |
+
"isHybrid": true,
|
| 832 |
+
"isExternalPureBaseline": false,
|
| 833 |
+
"isExternalRebuiltBaseline": false,
|
| 834 |
+
"isMaterializedTensorMapped": true,
|
| 835 |
+
"kld": 0.013794,
|
| 836 |
+
"ppl": 5.641085,
|
| 837 |
+
"pplDeltaPercent": 0.5989300044583223,
|
| 838 |
+
"sizeBytes": 16337626816,
|
| 839 |
+
"sizeGB": 16.337626816,
|
| 840 |
+
"sizeGiB": 15.215600669384003
|
| 841 |
+
},
|
| 842 |
+
"deltas": {
|
| 843 |
+
"kld": 0.0011979999999999994,
|
| 844 |
+
"sizeBytes": 0,
|
| 845 |
+
"sizeGB": 0,
|
| 846 |
+
"sizeGiB": 0,
|
| 847 |
+
"removedPplDeltaPercent": 0.8432278198840897,
|
| 848 |
+
"winnerPplDeltaPercent": 0.5989300044583223,
|
| 849 |
+
"pplDeltaPercentImprovement": 0.24429781542576734
|
| 850 |
+
}
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 854 |
+
"reasonDescription": "The winner used only the configured near-baseline size premium and beat the real linear KLD trade line.",
|
| 855 |
+
"rawReason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium",
|
| 856 |
+
"removed": {
|
| 857 |
+
"key": "107:0:0:0:0:0:0:0:0:0",
|
| 858 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-IQ4_XS.gguf",
|
| 859 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-IQ4_XS",
|
| 860 |
+
"shortName": "Unsloth-IQ4_XS",
|
| 861 |
+
"provider": "Unsloth",
|
| 862 |
+
"quantFamily": "Unsloth-IQ4_XS",
|
| 863 |
+
"isHybrid": false,
|
| 864 |
+
"isExternalPureBaseline": true,
|
| 865 |
+
"isExternalRebuiltBaseline": false,
|
| 866 |
+
"isMaterializedTensorMapped": false,
|
| 867 |
+
"kld": 0.015238,
|
| 868 |
+
"ppl": 5.666636,
|
| 869 |
+
"pplDeltaPercent": 1.054587605884969,
|
| 870 |
+
"sizeBytes": 15705859776,
|
| 871 |
+
"sizeGB": 15.705859776,
|
| 872 |
+
"sizeGiB": 14.627221763134003
|
| 873 |
+
},
|
| 874 |
+
"winner": {
|
| 875 |
+
"key": "107:108:108:110:115:118:108:108:0:0",
|
| 876 |
+
"fileName": "Qwen3.8-27B-MQ-IQ4_XS.gguf",
|
| 877 |
+
"displayName": "Qwen3.8-27B-MQ-IQ4_XS",
|
| 878 |
+
"shortName": "MQ-Unsloth-IQ4_XS",
|
| 879 |
+
"provider": "MagicQuant",
|
| 880 |
+
"quantFamily": "Unsloth-IQ4_XS",
|
| 881 |
+
"isHybrid": true,
|
| 882 |
+
"isExternalPureBaseline": false,
|
| 883 |
+
"isExternalRebuiltBaseline": false,
|
| 884 |
+
"isMaterializedTensorMapped": true,
|
| 885 |
+
"kld": 0.014337,
|
| 886 |
+
"ppl": 5.650981,
|
| 887 |
+
"pplDeltaPercent": 0.7754079358002655,
|
| 888 |
+
"sizeBytes": 15904679616,
|
| 889 |
+
"sizeGB": 15.904679616000001,
|
| 890 |
+
"sizeGiB": 14.81238716840744
|
| 891 |
+
},
|
| 892 |
+
"deltas": {
|
| 893 |
+
"kld": 0.000900999999999999,
|
| 894 |
+
"sizeBytes": -198819840,
|
| 895 |
+
"sizeGB": -0.19881984,
|
| 896 |
+
"sizeGiB": -0.1851654052734375,
|
| 897 |
+
"removedPplDeltaPercent": 1.054587605884969,
|
| 898 |
+
"winnerPplDeltaPercent": 0.7754079358002655,
|
| 899 |
+
"pplDeltaPercentImprovement": 0.27917967008470357
|
| 900 |
+
}
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 904 |
+
"reasonDescription": "Two candidates were too close in practical output space; the stronger one was kept.",
|
| 905 |
+
"rawReason": "meaningful spacing collapse; size gap below 599,772,979 bytes",
|
| 906 |
+
"removed": {
|
| 907 |
+
"key": "107:0:0:0:0:0:0:0:0:0",
|
| 908 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-IQ4_XS.gguf",
|
| 909 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-IQ4_XS",
|
| 910 |
+
"shortName": "Unsloth-IQ4_XS",
|
| 911 |
+
"provider": "Unsloth",
|
| 912 |
+
"quantFamily": "Unsloth-IQ4_XS",
|
| 913 |
+
"isHybrid": false,
|
| 914 |
+
"isExternalPureBaseline": true,
|
| 915 |
+
"isExternalRebuiltBaseline": false,
|
| 916 |
+
"isMaterializedTensorMapped": false,
|
| 917 |
+
"kld": 0.015238,
|
| 918 |
+
"ppl": 5.666636,
|
| 919 |
+
"pplDeltaPercent": 1.054587605884969,
|
| 920 |
+
"sizeBytes": 15705859776,
|
| 921 |
+
"sizeGB": 15.705859776,
|
| 922 |
+
"sizeGiB": 14.627221763134003
|
| 923 |
+
},
|
| 924 |
+
"winner": {
|
| 925 |
+
"key": "107:108:108:110:115:118:108:108:0:0",
|
| 926 |
+
"fileName": "Qwen3.8-27B-MQ-IQ4_XS.gguf",
|
| 927 |
+
"displayName": "Qwen3.8-27B-MQ-IQ4_XS",
|
| 928 |
+
"shortName": "MQ-Unsloth-IQ4_XS",
|
| 929 |
+
"provider": "MagicQuant",
|
| 930 |
+
"quantFamily": "Unsloth-IQ4_XS",
|
| 931 |
+
"isHybrid": true,
|
| 932 |
+
"isExternalPureBaseline": false,
|
| 933 |
+
"isExternalRebuiltBaseline": false,
|
| 934 |
+
"isMaterializedTensorMapped": true,
|
| 935 |
+
"kld": 0.014337,
|
| 936 |
+
"ppl": 5.650981,
|
| 937 |
+
"pplDeltaPercent": 0.7754079358002655,
|
| 938 |
+
"sizeBytes": 15904679616,
|
| 939 |
+
"sizeGB": 15.904679616000001,
|
| 940 |
+
"sizeGiB": 14.81238716840744
|
| 941 |
+
},
|
| 942 |
+
"deltas": {
|
| 943 |
+
"kld": 0.000900999999999999,
|
| 944 |
+
"sizeBytes": -198819840,
|
| 945 |
+
"sizeGB": -0.19881984,
|
| 946 |
+
"sizeGiB": -0.1851654052734375,
|
| 947 |
+
"removedPplDeltaPercent": 1.054587605884969,
|
| 948 |
+
"winnerPplDeltaPercent": 0.7754079358002655,
|
| 949 |
+
"pplDeltaPercentImprovement": 0.27917967008470357
|
| 950 |
+
}
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 954 |
+
"reasonDescription": "The winner used only the configured near-baseline size premium and beat the real linear KLD trade line.",
|
| 955 |
+
"rawReason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium",
|
| 956 |
+
"removed": {
|
| 957 |
+
"key": "103:0:0:0:0:0:0:0:0:0",
|
| 958 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-UD-IQ3_XXS.gguf",
|
| 959 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-UD-IQ3_XXS",
|
| 960 |
+
"shortName": "Unsloth-UD-IQ3_XXS",
|
| 961 |
+
"provider": "Unsloth",
|
| 962 |
+
"quantFamily": "Unsloth-UD-IQ3_XXS",
|
| 963 |
+
"isHybrid": false,
|
| 964 |
+
"isExternalPureBaseline": true,
|
| 965 |
+
"isExternalRebuiltBaseline": false,
|
| 966 |
+
"isMaterializedTensorMapped": false,
|
| 967 |
+
"kld": 0.058766,
|
| 968 |
+
"ppl": 5.876035,
|
| 969 |
+
"pplDeltaPercent": 4.788854213107445,
|
| 970 |
+
"sizeBytes": 11954476736,
|
| 971 |
+
"sizeGB": 11.954476735999998,
|
| 972 |
+
"sizeGiB": 11.133474051952362
|
| 973 |
+
},
|
| 974 |
+
"winner": {
|
| 975 |
+
"key": "103:104:104:104:109:104:104:104:0:0",
|
| 976 |
+
"fileName": "Qwen3.8-27B-MQ-IQ2_M_1.gguf",
|
| 977 |
+
"displayName": "MQ-IQ2_M_1",
|
| 978 |
+
"shortName": "MQ-IQ2_M_1",
|
| 979 |
+
"provider": "MagicQuant",
|
| 980 |
+
"quantFamily": "IQ2_M",
|
| 981 |
+
"isHybrid": true,
|
| 982 |
+
"isExternalPureBaseline": false,
|
| 983 |
+
"isExternalRebuiltBaseline": false,
|
| 984 |
+
"isMaterializedTensorMapped": true,
|
| 985 |
+
"kld": 0.057811,
|
| 986 |
+
"ppl": 5.8607,
|
| 987 |
+
"pplDeltaPercent": 4.5153811859117186,
|
| 988 |
+
"sizeBytes": 11960702656,
|
| 989 |
+
"sizeGB": 11.960702655999999,
|
| 990 |
+
"sizeGiB": 11.139272391796112
|
| 991 |
+
},
|
| 992 |
+
"deltas": {
|
| 993 |
+
"kld": 0.0009549999999999975,
|
| 994 |
+
"sizeBytes": -6225920,
|
| 995 |
+
"sizeGB": -0.00622592,
|
| 996 |
+
"sizeGiB": -0.00579833984375,
|
| 997 |
+
"removedPplDeltaPercent": 4.788854213107445,
|
| 998 |
+
"winnerPplDeltaPercent": 4.5153811859117186,
|
| 999 |
+
"pplDeltaPercentImprovement": 0.27347302719572664
|
| 1000 |
+
}
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 1004 |
+
"reasonDescription": "Two candidates were too close in practical output space; the stronger one was kept.",
|
| 1005 |
+
"rawReason": "meaningful spacing collapse; size gap below 599,772,979 bytes",
|
| 1006 |
+
"removed": {
|
| 1007 |
+
"key": "103:0:0:0:0:0:0:0:0:0",
|
| 1008 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-UD-IQ3_XXS.gguf",
|
| 1009 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-UD-IQ3_XXS",
|
| 1010 |
+
"shortName": "Unsloth-UD-IQ3_XXS",
|
| 1011 |
+
"provider": "Unsloth",
|
| 1012 |
+
"quantFamily": "Unsloth-UD-IQ3_XXS",
|
| 1013 |
+
"isHybrid": false,
|
| 1014 |
+
"isExternalPureBaseline": true,
|
| 1015 |
+
"isExternalRebuiltBaseline": false,
|
| 1016 |
+
"isMaterializedTensorMapped": false,
|
| 1017 |
+
"kld": 0.058766,
|
| 1018 |
+
"ppl": 5.876035,
|
| 1019 |
+
"pplDeltaPercent": 4.788854213107445,
|
| 1020 |
+
"sizeBytes": 11954476736,
|
| 1021 |
+
"sizeGB": 11.954476735999998,
|
| 1022 |
+
"sizeGiB": 11.133474051952362
|
| 1023 |
+
},
|
| 1024 |
+
"winner": {
|
| 1025 |
+
"key": "103:104:104:104:109:104:104:104:0:0",
|
| 1026 |
+
"fileName": "Qwen3.8-27B-MQ-IQ2_M_1.gguf",
|
| 1027 |
+
"displayName": "MQ-IQ2_M_1",
|
| 1028 |
+
"shortName": "MQ-IQ2_M_1",
|
| 1029 |
+
"provider": "MagicQuant",
|
| 1030 |
+
"quantFamily": "IQ2_M",
|
| 1031 |
+
"isHybrid": true,
|
| 1032 |
+
"isExternalPureBaseline": false,
|
| 1033 |
+
"isExternalRebuiltBaseline": false,
|
| 1034 |
+
"isMaterializedTensorMapped": true,
|
| 1035 |
+
"kld": 0.057811,
|
| 1036 |
+
"ppl": 5.8607,
|
| 1037 |
+
"pplDeltaPercent": 4.5153811859117186,
|
| 1038 |
+
"sizeBytes": 11960702656,
|
| 1039 |
+
"sizeGB": 11.960702655999999,
|
| 1040 |
+
"sizeGiB": 11.139272391796112
|
| 1041 |
+
},
|
| 1042 |
+
"deltas": {
|
| 1043 |
+
"kld": 0.0009549999999999975,
|
| 1044 |
+
"sizeBytes": -6225920,
|
| 1045 |
+
"sizeGB": -0.00622592,
|
| 1046 |
+
"sizeGiB": -0.00579833984375,
|
| 1047 |
+
"removedPplDeltaPercent": 4.788854213107445,
|
| 1048 |
+
"winnerPplDeltaPercent": 4.5153811859117186,
|
| 1049 |
+
"pplDeltaPercentImprovement": 0.27347302719572664
|
| 1050 |
+
}
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 1054 |
+
"reasonDescription": "The winner used only the configured near-baseline size premium and beat the real linear KLD trade line.",
|
| 1055 |
+
"rawReason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium",
|
| 1056 |
+
"removed": {
|
| 1057 |
+
"key": "102:0:0:0:0:0:0:0:0:0",
|
| 1058 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-UD-Q2_K_XL.gguf",
|
| 1059 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-UD-Q2_K_XL",
|
| 1060 |
+
"shortName": "Unsloth-UD-Q2_K_XL",
|
| 1061 |
+
"provider": "Unsloth",
|
| 1062 |
+
"quantFamily": "Unsloth-UD-Q2_K_XL",
|
| 1063 |
+
"isHybrid": false,
|
| 1064 |
+
"isExternalPureBaseline": true,
|
| 1065 |
+
"isExternalRebuiltBaseline": false,
|
| 1066 |
+
"isMaterializedTensorMapped": false,
|
| 1067 |
+
"kld": 0.094036,
|
| 1068 |
+
"ppl": 6.019528,
|
| 1069 |
+
"pplDeltaPercent": 7.347802050824794,
|
| 1070 |
+
"sizeBytes": 10721027776,
|
| 1071 |
+
"sizeGB": 10.721027776000001,
|
| 1072 |
+
"sizeGiB": 9.984735190868378
|
| 1073 |
+
},
|
| 1074 |
+
"winner": {
|
| 1075 |
+
"key": "102:103:103:103:114:103:103:103:0:0",
|
| 1076 |
+
"fileName": "Qwen3.8-27B-MQ-IQ2_M_2.gguf",
|
| 1077 |
+
"displayName": "MQ-IQ2_M_2",
|
| 1078 |
+
"shortName": "MQ-IQ2_M_2",
|
| 1079 |
+
"provider": "MagicQuant",
|
| 1080 |
+
"quantFamily": "IQ2_M",
|
| 1081 |
+
"isHybrid": true,
|
| 1082 |
+
"isExternalPureBaseline": false,
|
| 1083 |
+
"isExternalRebuiltBaseline": false,
|
| 1084 |
+
"isMaterializedTensorMapped": true,
|
| 1085 |
+
"kld": 0.092394,
|
| 1086 |
+
"ppl": 6.009092,
|
| 1087 |
+
"pplDeltaPercent": 7.161694159607668,
|
| 1088 |
+
"sizeBytes": 10769770176,
|
| 1089 |
+
"sizeGB": 10.769770176000002,
|
| 1090 |
+
"sizeGiB": 10.030130088329315
|
| 1091 |
+
},
|
| 1092 |
+
"deltas": {
|
| 1093 |
+
"kld": 0.0016419999999999907,
|
| 1094 |
+
"sizeBytes": -48742400,
|
| 1095 |
+
"sizeGB": -0.048742400000000005,
|
| 1096 |
+
"sizeGiB": -0.0453948974609375,
|
| 1097 |
+
"removedPplDeltaPercent": 7.347802050824794,
|
| 1098 |
+
"winnerPplDeltaPercent": 7.161694159607668,
|
| 1099 |
+
"pplDeltaPercentImprovement": 0.18610789121712568
|
| 1100 |
+
}
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 1104 |
+
"reasonDescription": "Two candidates were too close in practical output space; the stronger one was kept.",
|
| 1105 |
+
"rawReason": "meaningful spacing collapse; size gap below 599,772,979 bytes",
|
| 1106 |
+
"removed": {
|
| 1107 |
+
"key": "102:0:0:0:0:0:0:0:0:0",
|
| 1108 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-UD-Q2_K_XL.gguf",
|
| 1109 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-UD-Q2_K_XL",
|
| 1110 |
+
"shortName": "Unsloth-UD-Q2_K_XL",
|
| 1111 |
+
"provider": "Unsloth",
|
| 1112 |
+
"quantFamily": "Unsloth-UD-Q2_K_XL",
|
| 1113 |
+
"isHybrid": false,
|
| 1114 |
+
"isExternalPureBaseline": true,
|
| 1115 |
+
"isExternalRebuiltBaseline": false,
|
| 1116 |
+
"isMaterializedTensorMapped": false,
|
| 1117 |
+
"kld": 0.094036,
|
| 1118 |
+
"ppl": 6.019528,
|
| 1119 |
+
"pplDeltaPercent": 7.347802050824794,
|
| 1120 |
+
"sizeBytes": 10721027776,
|
| 1121 |
+
"sizeGB": 10.721027776000001,
|
| 1122 |
+
"sizeGiB": 9.984735190868378
|
| 1123 |
+
},
|
| 1124 |
+
"winner": {
|
| 1125 |
+
"key": "102:103:103:103:114:103:103:103:0:0",
|
| 1126 |
+
"fileName": "Qwen3.8-27B-MQ-IQ2_M_2.gguf",
|
| 1127 |
+
"displayName": "MQ-IQ2_M_2",
|
| 1128 |
+
"shortName": "MQ-IQ2_M_2",
|
| 1129 |
+
"provider": "MagicQuant",
|
| 1130 |
+
"quantFamily": "IQ2_M",
|
| 1131 |
+
"isHybrid": true,
|
| 1132 |
+
"isExternalPureBaseline": false,
|
| 1133 |
+
"isExternalRebuiltBaseline": false,
|
| 1134 |
+
"isMaterializedTensorMapped": true,
|
| 1135 |
+
"kld": 0.092394,
|
| 1136 |
+
"ppl": 6.009092,
|
| 1137 |
+
"pplDeltaPercent": 7.161694159607668,
|
| 1138 |
+
"sizeBytes": 10769770176,
|
| 1139 |
+
"sizeGB": 10.769770176000002,
|
| 1140 |
+
"sizeGiB": 10.030130088329315
|
| 1141 |
+
},
|
| 1142 |
+
"deltas": {
|
| 1143 |
+
"kld": 0.0016419999999999907,
|
| 1144 |
+
"sizeBytes": -48742400,
|
| 1145 |
+
"sizeGB": -0.048742400000000005,
|
| 1146 |
+
"sizeGiB": -0.0453948974609375,
|
| 1147 |
+
"removedPplDeltaPercent": 7.347802050824794,
|
| 1148 |
+
"winnerPplDeltaPercent": 7.161694159607668,
|
| 1149 |
+
"pplDeltaPercentImprovement": 0.18610789121712568
|
| 1150 |
+
}
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 1154 |
+
"reasonDescription": "Two candidates were too close in practical output space; the stronger one was kept.",
|
| 1155 |
+
"rawReason": "meaningful spacing collapse; size gap below 599,772,979 bytes",
|
| 1156 |
+
"removed": {
|
| 1157 |
+
"key": "101:102:102:102:113:102:102:102:0:0",
|
| 1158 |
+
"fileName": "Qwen3.8-27B-MQ-IQ2_M.gguf",
|
| 1159 |
+
"displayName": "Qwen3.8-27B-MQ-IQ2_M",
|
| 1160 |
+
"shortName": "MQ-Unsloth-UD-IQ2_M",
|
| 1161 |
+
"provider": "MagicQuant",
|
| 1162 |
+
"quantFamily": "Unsloth-UD-IQ2_M",
|
| 1163 |
+
"isHybrid": true,
|
| 1164 |
+
"isExternalPureBaseline": false,
|
| 1165 |
+
"isExternalRebuiltBaseline": false,
|
| 1166 |
+
"isMaterializedTensorMapped": true,
|
| 1167 |
+
"kld": 0.108218,
|
| 1168 |
+
"ppl": 6.097877,
|
| 1169 |
+
"pplDeltaPercent": 8.745020062416415,
|
| 1170 |
+
"sizeBytes": 10391299776,
|
| 1171 |
+
"sizeGB": 10.391299776,
|
| 1172 |
+
"sizeGiB": 9.677652060985565
|
| 1173 |
+
},
|
| 1174 |
+
"winner": {
|
| 1175 |
+
"key": "102:0:0:0:0:0:0:0:0:0",
|
| 1176 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-UD-Q2_K_XL.gguf",
|
| 1177 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-UD-Q2_K_XL",
|
| 1178 |
+
"shortName": "Unsloth-UD-Q2_K_XL",
|
| 1179 |
+
"provider": "Unsloth",
|
| 1180 |
+
"quantFamily": "Unsloth-UD-Q2_K_XL",
|
| 1181 |
+
"isHybrid": false,
|
| 1182 |
+
"isExternalPureBaseline": true,
|
| 1183 |
+
"isExternalRebuiltBaseline": false,
|
| 1184 |
+
"isMaterializedTensorMapped": false,
|
| 1185 |
+
"kld": 0.094036,
|
| 1186 |
+
"ppl": 6.019528,
|
| 1187 |
+
"pplDeltaPercent": 7.347802050824794,
|
| 1188 |
+
"sizeBytes": 10721027776,
|
| 1189 |
+
"sizeGB": 10.721027776000001,
|
| 1190 |
+
"sizeGiB": 9.984735190868378
|
| 1191 |
+
},
|
| 1192 |
+
"deltas": {
|
| 1193 |
+
"kld": 0.014182,
|
| 1194 |
+
"sizeBytes": -329728000,
|
| 1195 |
+
"sizeGB": -0.329728,
|
| 1196 |
+
"sizeGiB": -0.3070831298828125,
|
| 1197 |
+
"removedPplDeltaPercent": 8.745020062416415,
|
| 1198 |
+
"winnerPplDeltaPercent": 7.347802050824794,
|
| 1199 |
+
"pplDeltaPercentImprovement": 1.3972180115916215
|
| 1200 |
+
}
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"reasonCode": "NEAR_BASELINE_PREMIUM",
|
| 1204 |
+
"reasonDescription": "The winner used only the configured near-baseline size premium and beat the real linear KLD trade line.",
|
| 1205 |
+
"rawReason": "smart baseline-tuning near-baseline fallback within \u002B1.5% size premium",
|
| 1206 |
+
"removed": {
|
| 1207 |
+
"key": "101:0:0:0:0:0:0:0:0:0",
|
| 1208 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-UD-IQ2_M.gguf",
|
| 1209 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-UD-IQ2_M",
|
| 1210 |
+
"shortName": "Unsloth-UD-IQ2_M",
|
| 1211 |
+
"provider": "Unsloth",
|
| 1212 |
+
"quantFamily": "Unsloth-UD-IQ2_M",
|
| 1213 |
+
"isHybrid": false,
|
| 1214 |
+
"isExternalPureBaseline": true,
|
| 1215 |
+
"isExternalRebuiltBaseline": false,
|
| 1216 |
+
"isMaterializedTensorMapped": false,
|
| 1217 |
+
"kld": 0.109431,
|
| 1218 |
+
"ppl": 6.109936,
|
| 1219 |
+
"pplDeltaPercent": 8.960071333036119,
|
| 1220 |
+
"sizeBytes": 10364511936,
|
| 1221 |
+
"sizeGB": 10.364511936000001,
|
| 1222 |
+
"sizeGiB": 9.652703940868378
|
| 1223 |
+
},
|
| 1224 |
+
"winner": {
|
| 1225 |
+
"key": "101:102:102:102:113:102:102:102:0:0",
|
| 1226 |
+
"fileName": "Qwen3.8-27B-MQ-IQ2_M.gguf",
|
| 1227 |
+
"displayName": "Qwen3.8-27B-MQ-IQ2_M",
|
| 1228 |
+
"shortName": "MQ-Unsloth-UD-IQ2_M",
|
| 1229 |
+
"provider": "MagicQuant",
|
| 1230 |
+
"quantFamily": "Unsloth-UD-IQ2_M",
|
| 1231 |
+
"isHybrid": true,
|
| 1232 |
+
"isExternalPureBaseline": false,
|
| 1233 |
+
"isExternalRebuiltBaseline": false,
|
| 1234 |
+
"isMaterializedTensorMapped": true,
|
| 1235 |
+
"kld": 0.108218,
|
| 1236 |
+
"ppl": 6.097877,
|
| 1237 |
+
"pplDeltaPercent": 8.745020062416415,
|
| 1238 |
+
"sizeBytes": 10391299776,
|
| 1239 |
+
"sizeGB": 10.391299776,
|
| 1240 |
+
"sizeGiB": 9.677652060985565
|
| 1241 |
+
},
|
| 1242 |
+
"deltas": {
|
| 1243 |
+
"kld": 0.0012130000000000057,
|
| 1244 |
+
"sizeBytes": -26787840,
|
| 1245 |
+
"sizeGB": -0.02678784,
|
| 1246 |
+
"sizeGiB": -0.0249481201171875,
|
| 1247 |
+
"removedPplDeltaPercent": 8.960071333036119,
|
| 1248 |
+
"winnerPplDeltaPercent": 8.745020062416415,
|
| 1249 |
+
"pplDeltaPercentImprovement": 0.21505127061970342
|
| 1250 |
+
}
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"reasonCode": "SPACING_COLLAPSE",
|
| 1254 |
+
"reasonDescription": "Two candidates were too close in practical output space; the stronger one was kept.",
|
| 1255 |
+
"rawReason": "meaningful spacing collapse; size gap below 599,772,979 bytes",
|
| 1256 |
+
"removed": {
|
| 1257 |
+
"key": "101:0:0:0:0:0:0:0:0:0",
|
| 1258 |
+
"fileName": "Qwen3.8-27B-UD-Unsloth-UD-IQ2_M.gguf",
|
| 1259 |
+
"displayName": "Qwen3.8-27B-UD-Unsloth-UD-IQ2_M",
|
| 1260 |
+
"shortName": "Unsloth-UD-IQ2_M",
|
| 1261 |
+
"provider": "Unsloth",
|
| 1262 |
+
"quantFamily": "Unsloth-UD-IQ2_M",
|
| 1263 |
+
"isHybrid": false,
|
| 1264 |
+
"isExternalPureBaseline": true,
|
| 1265 |
+
"isExternalRebuiltBaseline": false,
|
| 1266 |
+
"isMaterializedTensorMapped": false,
|
| 1267 |
+
"kld": 0.109431,
|
| 1268 |
+
"ppl": 6.109936,
|
| 1269 |
+
"pplDeltaPercent": 8.960071333036119,
|
| 1270 |
+
"sizeBytes": 10364511936,
|
| 1271 |
+
"sizeGB": 10.364511936000001,
|
| 1272 |
+
"sizeGiB": 9.652703940868378
|
| 1273 |
+
},
|
| 1274 |
+
"winner": {
|
| 1275 |
+
"key": "101:102:102:102:113:102:102:102:0:0",
|
| 1276 |
+
"fileName": "Qwen3.8-27B-MQ-IQ2_M.gguf",
|
| 1277 |
+
"displayName": "Qwen3.8-27B-MQ-IQ2_M",
|
| 1278 |
+
"shortName": "MQ-Unsloth-UD-IQ2_M",
|
| 1279 |
+
"provider": "MagicQuant",
|
| 1280 |
+
"quantFamily": "Unsloth-UD-IQ2_M",
|
| 1281 |
+
"isHybrid": true,
|
| 1282 |
+
"isExternalPureBaseline": false,
|
| 1283 |
+
"isExternalRebuiltBaseline": false,
|
| 1284 |
+
"isMaterializedTensorMapped": true,
|
| 1285 |
+
"kld": 0.108218,
|
| 1286 |
+
"ppl": 6.097877,
|
| 1287 |
+
"pplDeltaPercent": 8.745020062416415,
|
| 1288 |
+
"sizeBytes": 10391299776,
|
| 1289 |
+
"sizeGB": 10.391299776,
|
| 1290 |
+
"sizeGiB": 9.677652060985565
|
| 1291 |
+
},
|
| 1292 |
+
"deltas": {
|
| 1293 |
+
"kld": 0.0012130000000000057,
|
| 1294 |
+
"sizeBytes": -26787840,
|
| 1295 |
+
"sizeGB": -0.02678784,
|
| 1296 |
+
"sizeGiB": -0.0249481201171875,
|
| 1297 |
+
"removedPplDeltaPercent": 8.960071333036119,
|
| 1298 |
+
"winnerPplDeltaPercent": 8.745020062416415,
|
| 1299 |
+
"pplDeltaPercentImprovement": 0.21505127061970342
|
| 1300 |
+
}
|
| 1301 |
+
}
|
| 1302 |
+
]
|
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
mmproj-BF16.gguf
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d1be8549eb05443ff965c3a054492865006e3ea9e5bc95fca4ae2de535ef2c01
|
| 3 |
+
size 927607040
|
model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0997f410c57a1f4e53b09e4be8f4a172d90edd9564368fb0847030937229b9f3
|
| 3 |
+
size 12809320
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,305 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"added_tokens_decoder": {
|
| 4 |
+
"248044": {
|
| 5 |
+
"content": "<|endoftext|>",
|
| 6 |
+
"lstrip": false,
|
| 7 |
+
"normalized": false,
|
| 8 |
+
"rstrip": false,
|
| 9 |
+
"single_word": false,
|
| 10 |
+
"special": true
|
| 11 |
+
},
|
| 12 |
+
"248045": {
|
| 13 |
+
"content": "<|im_start|>",
|
| 14 |
+
"lstrip": false,
|
| 15 |
+
"normalized": false,
|
| 16 |
+
"rstrip": false,
|
| 17 |
+
"single_word": false,
|
| 18 |
+
"special": true
|
| 19 |
+
},
|
| 20 |
+
"248046": {
|
| 21 |
+
"content": "<|im_end|>",
|
| 22 |
+
"lstrip": false,
|
| 23 |
+
"normalized": false,
|
| 24 |
+
"rstrip": false,
|
| 25 |
+
"single_word": false,
|
| 26 |
+
"special": true
|
| 27 |
+
},
|
| 28 |
+
"248047": {
|
| 29 |
+
"content": "<|object_ref_start|>",
|
| 30 |
+
"lstrip": false,
|
| 31 |
+
"normalized": false,
|
| 32 |
+
"rstrip": false,
|
| 33 |
+
"single_word": false,
|
| 34 |
+
"special": true
|
| 35 |
+
},
|
| 36 |
+
"248048": {
|
| 37 |
+
"content": "<|object_ref_end|>",
|
| 38 |
+
"lstrip": false,
|
| 39 |
+
"normalized": false,
|
| 40 |
+
"rstrip": false,
|
| 41 |
+
"single_word": false,
|
| 42 |
+
"special": true
|
| 43 |
+
},
|
| 44 |
+
"248049": {
|
| 45 |
+
"content": "<|box_start|>",
|
| 46 |
+
"lstrip": false,
|
| 47 |
+
"normalized": false,
|
| 48 |
+
"rstrip": false,
|
| 49 |
+
"single_word": false,
|
| 50 |
+
"special": true
|
| 51 |
+
},
|
| 52 |
+
"248050": {
|
| 53 |
+
"content": "<|box_end|>",
|
| 54 |
+
"lstrip": false,
|
| 55 |
+
"normalized": false,
|
| 56 |
+
"rstrip": false,
|
| 57 |
+
"single_word": false,
|
| 58 |
+
"special": true
|
| 59 |
+
},
|
| 60 |
+
"248051": {
|
| 61 |
+
"content": "<|quad_start|>",
|
| 62 |
+
"lstrip": false,
|
| 63 |
+
"normalized": false,
|
| 64 |
+
"rstrip": false,
|
| 65 |
+
"single_word": false,
|
| 66 |
+
"special": true
|
| 67 |
+
},
|
| 68 |
+
"248052": {
|
| 69 |
+
"content": "<|quad_end|>",
|
| 70 |
+
"lstrip": false,
|
| 71 |
+
"normalized": false,
|
| 72 |
+
"rstrip": false,
|
| 73 |
+
"single_word": false,
|
| 74 |
+
"special": true
|
| 75 |
+
},
|
| 76 |
+
"248053": {
|
| 77 |
+
"content": "<|vision_start|>",
|
| 78 |
+
"lstrip": false,
|
| 79 |
+
"normalized": false,
|
| 80 |
+
"rstrip": false,
|
| 81 |
+
"single_word": false,
|
| 82 |
+
"special": true
|
| 83 |
+
},
|
| 84 |
+
"248054": {
|
| 85 |
+
"content": "<|vision_end|>",
|
| 86 |
+
"lstrip": false,
|
| 87 |
+
"normalized": false,
|
| 88 |
+
"rstrip": false,
|
| 89 |
+
"single_word": false,
|
| 90 |
+
"special": true
|
| 91 |
+
},
|
| 92 |
+
"248055": {
|
| 93 |
+
"content": "<|vision_pad|>",
|
| 94 |
+
"lstrip": false,
|
| 95 |
+
"normalized": false,
|
| 96 |
+
"rstrip": false,
|
| 97 |
+
"single_word": false,
|
| 98 |
+
"special": true
|
| 99 |
+
},
|
| 100 |
+
"248056": {
|
| 101 |
+
"content": "<|image_pad|>",
|
| 102 |
+
"lstrip": false,
|
| 103 |
+
"normalized": false,
|
| 104 |
+
"rstrip": false,
|
| 105 |
+
"single_word": false,
|
| 106 |
+
"special": true
|
| 107 |
+
},
|
| 108 |
+
"248057": {
|
| 109 |
+
"content": "<|video_pad|>",
|
| 110 |
+
"lstrip": false,
|
| 111 |
+
"normalized": false,
|
| 112 |
+
"rstrip": false,
|
| 113 |
+
"single_word": false,
|
| 114 |
+
"special": true
|
| 115 |
+
},
|
| 116 |
+
"248058": {
|
| 117 |
+
"content": "<tool_call>",
|
| 118 |
+
"lstrip": false,
|
| 119 |
+
"normalized": false,
|
| 120 |
+
"rstrip": false,
|
| 121 |
+
"single_word": false,
|
| 122 |
+
"special": false
|
| 123 |
+
},
|
| 124 |
+
"248059": {
|
| 125 |
+
"content": "</tool_call>",
|
| 126 |
+
"lstrip": false,
|
| 127 |
+
"normalized": false,
|
| 128 |
+
"rstrip": false,
|
| 129 |
+
"single_word": false,
|
| 130 |
+
"special": false
|
| 131 |
+
},
|
| 132 |
+
"248060": {
|
| 133 |
+
"content": "<|fim_prefix|>",
|
| 134 |
+
"lstrip": false,
|
| 135 |
+
"normalized": false,
|
| 136 |
+
"rstrip": false,
|
| 137 |
+
"single_word": false,
|
| 138 |
+
"special": false
|
| 139 |
+
},
|
| 140 |
+
"248061": {
|
| 141 |
+
"content": "<|fim_middle|>",
|
| 142 |
+
"lstrip": false,
|
| 143 |
+
"normalized": false,
|
| 144 |
+
"rstrip": false,
|
| 145 |
+
"single_word": false,
|
| 146 |
+
"special": false
|
| 147 |
+
},
|
| 148 |
+
"248062": {
|
| 149 |
+
"content": "<|fim_suffix|>",
|
| 150 |
+
"lstrip": false,
|
| 151 |
+
"normalized": false,
|
| 152 |
+
"rstrip": false,
|
| 153 |
+
"single_word": false,
|
| 154 |
+
"special": false
|
| 155 |
+
},
|
| 156 |
+
"248063": {
|
| 157 |
+
"content": "<|fim_pad|>",
|
| 158 |
+
"lstrip": false,
|
| 159 |
+
"normalized": false,
|
| 160 |
+
"rstrip": false,
|
| 161 |
+
"single_word": false,
|
| 162 |
+
"special": false
|
| 163 |
+
},
|
| 164 |
+
"248064": {
|
| 165 |
+
"content": "<|repo_name|>",
|
| 166 |
+
"lstrip": false,
|
| 167 |
+
"normalized": false,
|
| 168 |
+
"rstrip": false,
|
| 169 |
+
"single_word": false,
|
| 170 |
+
"special": false
|
| 171 |
+
},
|
| 172 |
+
"248065": {
|
| 173 |
+
"content": "<|file_sep|>",
|
| 174 |
+
"lstrip": false,
|
| 175 |
+
"normalized": false,
|
| 176 |
+
"rstrip": false,
|
| 177 |
+
"single_word": false,
|
| 178 |
+
"special": false
|
| 179 |
+
},
|
| 180 |
+
"248066": {
|
| 181 |
+
"content": "<tool_response>",
|
| 182 |
+
"lstrip": false,
|
| 183 |
+
"normalized": false,
|
| 184 |
+
"rstrip": false,
|
| 185 |
+
"single_word": false,
|
| 186 |
+
"special": false
|
| 187 |
+
},
|
| 188 |
+
"248067": {
|
| 189 |
+
"content": "</tool_response>",
|
| 190 |
+
"lstrip": false,
|
| 191 |
+
"normalized": false,
|
| 192 |
+
"rstrip": false,
|
| 193 |
+
"single_word": false,
|
| 194 |
+
"special": false
|
| 195 |
+
},
|
| 196 |
+
"248068": {
|
| 197 |
+
"content": "<think>",
|
| 198 |
+
"lstrip": false,
|
| 199 |
+
"normalized": false,
|
| 200 |
+
"rstrip": false,
|
| 201 |
+
"single_word": false,
|
| 202 |
+
"special": false
|
| 203 |
+
},
|
| 204 |
+
"248069": {
|
| 205 |
+
"content": "</think>",
|
| 206 |
+
"lstrip": false,
|
| 207 |
+
"normalized": false,
|
| 208 |
+
"rstrip": false,
|
| 209 |
+
"single_word": false,
|
| 210 |
+
"special": false
|
| 211 |
+
},
|
| 212 |
+
"248070": {
|
| 213 |
+
"content": "<|audio_start|>",
|
| 214 |
+
"lstrip": false,
|
| 215 |
+
"normalized": false,
|
| 216 |
+
"rstrip": false,
|
| 217 |
+
"single_word": false,
|
| 218 |
+
"special": true
|
| 219 |
+
},
|
| 220 |
+
"248071": {
|
| 221 |
+
"content": "<|audio_end|>",
|
| 222 |
+
"lstrip": false,
|
| 223 |
+
"normalized": false,
|
| 224 |
+
"rstrip": false,
|
| 225 |
+
"single_word": false,
|
| 226 |
+
"special": true
|
| 227 |
+
},
|
| 228 |
+
"248072": {
|
| 229 |
+
"content": "<tts_pad>",
|
| 230 |
+
"lstrip": false,
|
| 231 |
+
"normalized": false,
|
| 232 |
+
"rstrip": false,
|
| 233 |
+
"single_word": false,
|
| 234 |
+
"special": true
|
| 235 |
+
},
|
| 236 |
+
"248073": {
|
| 237 |
+
"content": "<tts_text_bos>",
|
| 238 |
+
"lstrip": false,
|
| 239 |
+
"normalized": false,
|
| 240 |
+
"rstrip": false,
|
| 241 |
+
"single_word": false,
|
| 242 |
+
"special": true
|
| 243 |
+
},
|
| 244 |
+
"248074": {
|
| 245 |
+
"content": "<tts_text_eod>",
|
| 246 |
+
"lstrip": false,
|
| 247 |
+
"normalized": false,
|
| 248 |
+
"rstrip": false,
|
| 249 |
+
"single_word": false,
|
| 250 |
+
"special": true
|
| 251 |
+
},
|
| 252 |
+
"248075": {
|
| 253 |
+
"content": "<tts_text_bos_single>",
|
| 254 |
+
"lstrip": false,
|
| 255 |
+
"normalized": false,
|
| 256 |
+
"rstrip": false,
|
| 257 |
+
"single_word": false,
|
| 258 |
+
"special": true
|
| 259 |
+
},
|
| 260 |
+
"248076": {
|
| 261 |
+
"content": "<|audio_pad|>",
|
| 262 |
+
"lstrip": false,
|
| 263 |
+
"normalized": false,
|
| 264 |
+
"rstrip": false,
|
| 265 |
+
"single_word": false,
|
| 266 |
+
"special": true
|
| 267 |
+
}
|
| 268 |
+
},
|
| 269 |
+
"additional_special_tokens": [
|
| 270 |
+
"<|im_start|>",
|
| 271 |
+
"<|im_end|>",
|
| 272 |
+
"<|object_ref_start|>",
|
| 273 |
+
"<|object_ref_end|>",
|
| 274 |
+
"<|box_start|>",
|
| 275 |
+
"<|box_end|>",
|
| 276 |
+
"<|quad_start|>",
|
| 277 |
+
"<|quad_end|>",
|
| 278 |
+
"<|vision_start|>",
|
| 279 |
+
"<|vision_end|>",
|
| 280 |
+
"<|vision_pad|>",
|
| 281 |
+
"<|image_pad|>",
|
| 282 |
+
"<|video_pad|>"
|
| 283 |
+
],
|
| 284 |
+
"bos_token": null,
|
| 285 |
+
"chat_template": "{%- set image_count = namespace(value=0) %}\n{%- set video_count = namespace(value=0) %}\n{%- macro render_content(content, do_vision_count, is_system_content=false) %}\n {%- if content is string %}\n {{- content }}\n {%- elif content is iterable and content is not mapping %}\n {%- for item in content %}\n {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}\n {%- if is_system_content %}\n {{- raise_exception('System message cannot contain images.') }}\n {%- endif %}\n {%- if do_vision_count %}\n {%- set image_count.value = image_count.value + 1 %}\n {%- endif %}\n {%- if add_vision_id %}\n {{- 'Picture ' ~ image_count.value ~ ': ' }}\n {%- endif %}\n {{- '<|vision_start|><|image_pad|><|vision_end|>' }}\n {%- elif 'video' in item or item.type == 'video' %}\n {%- if is_system_content %}\n {{- raise_exception('System message cannot contain videos.') }}\n {%- endif %}\n {%- if do_vision_count %}\n {%- set video_count.value = video_count.value + 1 %}\n {%- endif %}\n {%- if add_vision_id %}\n {{- 'Video ' ~ video_count.value ~ ': ' }}\n {%- endif %}\n {{- '<|vision_start|><|video_pad|><|vision_end|>' }}\n {%- elif 'text' in item %}\n {{- item.text }}\n {%- else %}\n {{- raise_exception('Unexpected item type in content.') }}\n {%- endif %}\n {%- endfor %}\n {%- elif content is none or content is undefined %}\n {{- '' }}\n {%- else %}\n {{- raise_exception('Unexpected content type.') }}\n {%- endif %}\n{%- endmacro %}\n{%- if not messages %}\n {{- raise_exception('No messages provided.') }}\n{%- endif %}\n{%- set reasoning_instructions = '' %}\n{%- if enable_thinking is undefined or enable_thinking is true %}\n {%- set resolved_reasoning_effort = reasoning_effort|default('xhigh') %}\n {%- if resolved_reasoning_effort not in ('xhigh', 'medium', 'low') %}\n {{- raise_exception('Unexpected reasoning effort ' ~ reasoning_effort ~ '. Supported types are xhigh (default), medium, and low.') }}\n {%- endif %}\n {%- if resolved_reasoning_effort == 'xhigh' %}\n {%- set reasoning_instructions = 'Reasoning effort is set to xhigh. Please think carefully through the task, validate key assumptions, consider plausible alternatives, and prioritize correctness, consistency, and clarity in the final answer.' %}\n {%- elif resolved_reasoning_effort == 'low' %}\n {%- set reasoning_instructions = 'Reasoning effort is set to low. Keep your thinking brief and focused, moving directly to the conclusion without unnecessary elaboration.' %}\n {%- endif %}\n{%- endif %}\n{%- if tools and tools is iterable and tools is not mapping %}\n {{- '<|im_start|>system\\n' }}\n {%- if reasoning_instructions %}\n {{- reasoning_instructions + '\\n\\n' }}\n {%- endif %}\n {{- \"# Tools\\n\\nYou have access to the following functions:\\n\\n<tools>\" }}\n {%- for tool in tools %}\n {{- \"\\n\" }}\n {{- tool | tojson }}\n {%- endfor %}\n {{- \"\\n</tools>\" }}\n {{- '\\n\\nIf you choose to call a function ONLY reply in the following format with NO suffix:\\n\\n<tool_call>\\n<function=example_function_name>\\n<parameter=example_parameter_1>\\nvalue_1\\n</parameter>\\n<parameter=example_parameter_2>\\nThis is the value for the second parameter\\nthat can span\\nmultiple lines\\n</parameter>\\n</function>\\n</tool_call>\\n\\n<IMPORTANT>\\nReminder:\\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\\n- Required parameters MUST be specified\\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\\n</IMPORTANT>' }}\n {%- if messages[0].role == 'system' %}\n {%- set content = render_content(messages[0].content, false, true)|trim %}\n {%- if content %}\n {{- '\\n\\n' + content }}\n {%- endif %}\n {%- endif %}\n {{- '<|im_end|>\\n' }}\n{%- else %}\n {%- if messages[0].role == 'system' %}\n {%- set content = render_content(messages[0].content, false, true)|trim %}\n {%- if content %}\n {{- '<|im_start|>system\\n' + (reasoning_instructions + '\\n\\n' if reasoning_instructions else '') + content + '<|im_end|>\\n' }}\n {%- elif reasoning_instructions %}\n {{- '<|im_start|>system\\n' + reasoning_instructions + '<|im_end|>\\n' }}\n {%- endif %}\n {%- elif reasoning_instructions %}\n {{- '<|im_start|>system\\n' + reasoning_instructions + '<|im_end|>\\n' }}\n {%- endif %}\n{%- endif %}\n{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}\n{%- for message in messages[::-1] %}\n {%- set index = (messages|length - 1) - loop.index0 %}\n {%- if ns.multi_step_tool and message.role == \"user\" %}\n {%- set content = render_content(message.content, false)|trim %}\n {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}\n {%- set ns.multi_step_tool = false %}\n {%- set ns.last_query_index = index %}\n {%- endif %}\n {%- endif %}\n{%- endfor %}\n{%- if ns.multi_step_tool %}\n {{- raise_exception('No user query found in messages.') }}\n{%- endif %}\n{%- for message in messages %}\n {%- set content = render_content(message.content, true)|trim %}\n {%- if message.role == \"system\" %}\n {%- if not loop.first %}\n {{- raise_exception('System message must be at the beginning.') }}\n {%- endif %}\n {%- elif message.role == \"user\" %}\n {{- '<|im_start|>' + message.role + '\\n' + content + '<|im_end|>' + '\\n' }}\n {%- elif message.role == \"assistant\" %}\n {%- set reasoning_content = '' %}\n {%- if message.reasoning_content is string %}\n {%- set reasoning_content = message.reasoning_content %}\n {%- endif %}\n {%- set reasoning_content = reasoning_content|trim %}\n {%- if preserve_thinking is undefined or preserve_thinking is true or loop.index0 > ns.last_query_index %}\n {{- '<|im_start|>' + message.role + '\\n<think>\\n' + reasoning_content + '\\n</think>\\n\\n' + content }}\n {%- else %}\n {{- '<|im_start|>' + message.role + '\\n' + content }}\n {%- endif %}\n {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}\n {%- for tool_call in message.tool_calls %}\n {%- if tool_call.function is defined %}\n {%- set tool_call = tool_call.function %}\n {%- endif %}\n {%- if loop.first %}\n {%- if content|trim %}\n {{- '\\n\\n<tool_call>\\n<function=' + tool_call.name + '>\\n' }}\n {%- else %}\n {{- '<tool_call>\\n<function=' + tool_call.name + '>\\n' }}\n {%- endif %}\n {%- else %}\n {{- '\\n<tool_call>\\n<function=' + tool_call.name + '>\\n' }}\n {%- endif %}\n {%- if tool_call.arguments is defined and tool_call.arguments != '' %}\n {%- for args_name, args_value in tool_call.arguments|items %}\n {{- '<parameter=' + args_name + '>\\n' }}\n {%- set args_value = args_value | string if args_value is string else args_value | tojson | safe %}\n {{- args_value }}\n {{- '\\n</parameter>\\n' }}\n {%- endfor %}\n {%- endif %}\n {{- '</function>\\n</tool_call>' }}\n {%- endfor %}\n {%- endif %}\n {{- '<|im_end|>\\n' }}\n {%- elif message.role == \"tool\" %}\n {%- if loop.previtem and loop.previtem.role != \"tool\" %}\n {{- '<|im_start|>user' }}\n {%- endif %}\n {{- '\\n<tool_response>\\n' }}\n {{- content }}\n {{- '\\n</tool_response>' }}\n {%- if not loop.last and loop.nextitem.role != \"tool\" %}\n {{- '<|im_end|>\\n' }}\n {%- elif loop.last %}\n {{- '<|im_end|>\\n' }}\n {%- endif %}\n {%- else %}\n {{- raise_exception('Unexpected message role.') }}\n {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n {{- '<|im_start|>assistant\\n' }}\n {%- if enable_thinking is defined and enable_thinking is false %}\n {{- '<think>\\n\\n</think>\\n\\n' }}\n {%- else %}\n {{- '<think>\\n' }}\n {%- endif %}\n{%- endif %}",
|
| 286 |
+
"clean_up_tokenization_spaces": false,
|
| 287 |
+
"eos_token": "<|im_end|>",
|
| 288 |
+
"errors": "replace",
|
| 289 |
+
"model_max_length": 262144,
|
| 290 |
+
"pad_token": "<|endoftext|>",
|
| 291 |
+
"split_special_tokens": false,
|
| 292 |
+
"tokenizer_class": "Qwen2Tokenizer",
|
| 293 |
+
"unk_token": null,
|
| 294 |
+
"add_bos_token": false,
|
| 295 |
+
"pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
|
| 296 |
+
"extra_special_tokens": {
|
| 297 |
+
"audio_bos_token": "<|audio_start|>",
|
| 298 |
+
"audio_eos_token": "<|audio_end|>",
|
| 299 |
+
"audio_token": "<|audio_pad|>",
|
| 300 |
+
"image_token": "<|image_pad|>",
|
| 301 |
+
"video_token": "<|video_pad|>",
|
| 302 |
+
"vision_bos_token": "<|vision_start|>",
|
| 303 |
+
"vision_eos_token": "<|vision_end|>"
|
| 304 |
+
}
|
| 305 |
+
}
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|