Satinder Grewal 2969d6d15d model: add Hy3 (hy_v3) support with MTP speculative decoding (#25395) ... * model: add Hy3 (hy_v3) architecture support Adds Tencent Hunyuan 3 (HF architecture HYV3ForCausalLM, GGUF arch hy_v3): a MoE decoder stack with per-head Q/K RMSNorm, a sigmoid router with expert selection bias, an always-active ungated shared expert, and leading dense block(s) (first_k_dense_replace). The base implementation is ported from charlie12345's fork (https://github.com/charlie12345/ROCmFPX, src/models/hyv3.cpp), adapted to current mainline APIs (hparams.n_layer(), build_qkv, build_moe_ffn with fused gate_up + scale tensors, output_s). Note: blk.N.exp_probs_b is stored without a .bias suffix for compatibility with existing hy_v3 GGUFs produced by that fork. Co-Authored-By: charlie12345 <charlie12345@users.noreply.github.com> Co-authored-by: Piotr Wilkin <ilintar@gmail.com> Assisted-by: Claude Fable 5		2026-07-14 00:31:04 +02:00
..
templates	model: add Hy3 (hy_v3) support with MTP speculative decoding (#25395)	2026-07-14 00:31:04 +02:00
.editorconfig	gguf : new file format with flexible meta data (beta) (#2398)	2023-08-21 23:07:43 +03:00
ggml-vocab-aquila.gguf	Work on the BPE tokenizer (#3252)	2023-10-03 09:16:26 +02:00
ggml-vocab-baichuan.gguf	Add more tokenizer tests (#3742)	2023-10-24 09:17:17 +02:00
ggml-vocab-bert-bge.gguf	llama : fix BPE pre-tokenization (#6920)	2024-04-29 16:58:41 +03:00
ggml-vocab-bert-bge.gguf.inp	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-bert-bge.gguf.out	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-command-r.gguf	command-r : add BPE pre-tokenization (#7063)	2024-05-05 08:19:30 +03:00
ggml-vocab-command-r.gguf.inp	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-command-r.gguf.out	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-deepseek-coder.gguf	llama : fix BPE pre-tokenization (#6920)	2024-04-29 16:58:41 +03:00
ggml-vocab-deepseek-coder.gguf.inp	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-deepseek-coder.gguf.out	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-deepseek-llm.gguf	llama : fix BPE pre-tokenization (#6920)	2024-04-29 16:58:41 +03:00
ggml-vocab-deepseek-llm.gguf.inp	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-deepseek-llm.gguf.out	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-falcon.gguf	llama : fix BPE pre-tokenization (#6920)	2024-04-29 16:58:41 +03:00
ggml-vocab-falcon.gguf.inp	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-falcon.gguf.out	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-gemma-4.gguf	vocab: add gemma4 tokenizer tests, fix edge case (#21534)	2026-04-09 11:41:14 +02:00
ggml-vocab-gemma-4.gguf.inp	vocab: add gemma4 tokenizer tests, fix edge case (#21534)	2026-04-09 11:41:14 +02:00
ggml-vocab-gemma-4.gguf.out	vocab: add gemma4 tokenizer tests, fix edge case (#21534)	2026-04-09 11:41:14 +02:00
ggml-vocab-gpt-2.gguf	llama : fix BPE pre-tokenization (#6920)	2024-04-29 16:58:41 +03:00
ggml-vocab-gpt-2.gguf.inp	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-gpt-2.gguf.out	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-gpt-neox.gguf	Add more tokenizer tests (#3742)	2023-10-24 09:17:17 +02:00
ggml-vocab-llama-bpe.gguf	llama : fix BPE pre-tokenization (#6920)	2024-04-29 16:58:41 +03:00
ggml-vocab-llama-bpe.gguf.inp	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-llama-bpe.gguf.out	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-llama-spm.gguf	llama : fix BPE pre-tokenization (#6920)	2024-04-29 16:58:41 +03:00
ggml-vocab-llama-spm.gguf.inp	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-llama-spm.gguf.out	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-mpt.gguf	llama : fix BPE pre-tokenization (#6920)	2024-04-29 16:58:41 +03:00
ggml-vocab-mpt.gguf.inp	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-mpt.gguf.out	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-nomic-bert-moe.gguf	tests : improve UGM tokenizer test coverage (#13773)	2025-05-25 16:22:29 +02:00
ggml-vocab-phi-3.gguf	Per token attributes (#7685)	2024-06-04 09:17:17 +02:00
ggml-vocab-phi-3.gguf.inp	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-phi-3.gguf.out	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-qwen2.gguf	llama : add BPE pre-tokenization for Qwen2 (#7114)	2024-05-08 15:06:43 +03:00
ggml-vocab-qwen2.gguf.inp	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-qwen2.gguf.out	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-qwen35.gguf	unicode,test: add Qwen3.5 non-backtracking tokenizer handler and regr… (#22110)	2026-05-14 11:03:40 +02:00
ggml-vocab-qwen35.gguf.inp	unicode,test: add Qwen3.5 non-backtracking tokenizer handler and regr… (#22110)	2026-05-14 11:03:40 +02:00
ggml-vocab-qwen35.gguf.out	unicode,test: add Qwen3.5 non-backtracking tokenizer handler and regr… (#22110)	2026-05-14 11:03:40 +02:00
ggml-vocab-refact.gguf	tests : add test-tokenizer-0.sh + fix some tokenizers (#7036)	2024-05-04 08:32:32 +03:00
ggml-vocab-refact.gguf.inp	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-refact.gguf.out	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-starcoder.gguf	llama : fix BPE pre-tokenization (#6920)	2024-04-29 16:58:41 +03:00
ggml-vocab-starcoder.gguf.inp	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00
ggml-vocab-starcoder.gguf.out	convert : allow partial update to the chkhsh pre-tokenizer list (#13847)	2025-05-30 12:24:37 +02:00