File size: 2,557 Bytes
0985c38 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 | MISAKA PALW runtime for Qwen3.6-35B-A3B
Copyright 2026 MISAKA-BTC
This product is licensed under the Apache License, Version 2.0 (see LICENSE).
================================================================================
IMPORTANT — what this project is
================================================================================
This repository is a PROOF-OF-COMPUTE RUNTIME that executes an unmodified
upstream Qwen3.6-35B-A3B mixture-of-experts model as a local compute worker and
commits the execution as a signed `ComputeReceiptV1`. It does NOT train,
fine-tune, or otherwise modify the model, and it is NOT itself a new language
model. Model weights are not distributed in this repository; they are fetched at
pinned upstream artifacts (see config/runtime-pins.sh). Production Network
readiness is NOT claimed — see docs/requirements.md for the In-progress
requirements.
================================================================================
Third-party components
================================================================================
1. Qwen3.6-35B-A3B (huihui-ai abliterated release)
Base architecture Copyright Alibaba Cloud, licensed under the Apache License,
Version 2.0. Abliterated derivative by huihui-ai.
https://huggingface.co/huihui-ai/Huihui-Qwen3.6-35B-A3B-Claude-4.7-Opus-abliterated
https://ollama.com/huihui_ai/Qwen3.6-abliterated
Base metadata revision: ac18882735d037f6074a7630eb68d85db8234c25
GGUF artifact: Ollama huihui_ai/Qwen3.6-abliterated:35b-Claude-4.7,
blob sha256 1dc494614bee8a3bc00e79fe5a49da0fc1c36b3b118c4156e223e98e5a0a671b (Q4_K_M)
The upstream artifact is executed unmodified. Weights are fetched at the
pinned artifacts above and are not redistributed in this repository.
2. llama.cpp (ggml-org/llama.cpp)
Copyright (c) 2023-2026 The ggml authors
Licensed under the MIT License.
https://github.com/ggml-org/llama.cpp
Pinned at commit 12127defda4f41b7679cb2477a4b0d65ee6a0c8f.
patches/llama.cpp-palw-observer.patch redistributes MODIFICATIONS to llama.cpp
source (a read-only graph observer and qwen35moe loader/graph fixes required
to run the pinned GGUF) under the MIT License. The llama.cpp source itself is
not vendored in this repository; it is reconstructed from the pinned commit
(see docs/SETUP.md). The full MIT license text is included at
LICENSES/llama.cpp-MIT.txt.
Rust crate dependencies retain their own upstream licenses as declared in
runtime-palw/Cargo.toml and runtime-palw/Cargo.lock.
|