tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-07-16 15:47:35 +00:00
Code Issues Packages Projects Releases Wiki Activity

model : add hunyuan moe (#14425)

Browse Source 
* model : add hunyuan moe

* tokenizer ok

* fix tensor name

* cgraph init

* chat template

* wip

* almost working

* skip embed, fix bos

* cleanup

* yarn scaling

* cleanup

* correct rope type

* failed token fix

* ntk alpha freq_base

* tokenization working

* cleanup and pr changes

* vocab_size sanity check

* ntk alpha generic

* Update convert_hf_to_gguf.py

* Apply suggestions from code review

* fix regression

* fix style

---------

Co-authored-by: kooshi <1934337+kooshi@users.noreply.github.com>
This commit is contained in:
Xuan-Son Nguyen
2025-07-08 10:24:06 +02:00
committed by GitHub
parent 53903ae6fa
commit 8f22dc0a53
12 changed files with 449 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
gguf-py/gguf/tensor_mapping.py
View File

@ -303,6 +303,7 @@ class TensorNameMap:
"model.layers.{bid}.block_sparse_moe.router.layer", # granitemoe
"model.layers.{bid}.feed_forward.router", # llama4
"encoder.layers.{bid}.mlp.router.layer", # nomic-bert-moe
"model.layers.{bid}.mlp.gate.wg", # hunyuan
),
MODEL_TENSOR.FFN_GATE_INP_SHEXP: (
@ -362,6 +363,7 @@ class TensorNameMap:
"model.layers.{bid}.mlp.shared_expert.up_proj", # qwen2moe
"model.layers.{bid}.mlp.shared_experts.up_proj", # deepseek deepseek2
"model.layers.{bid}.feed_forward.shared_expert.up_proj", # llama4
"model.layers.{bid}.mlp.shared_mlp.up_proj", # hunyuan
),
# AWQ-activation gate
@ -398,6 +400,7 @@ class TensorNameMap:
"model.layers.{bid}.mlp.shared_expert.gate_proj", # qwen2moe
"model.layers.{bid}.mlp.shared_experts.gate_proj", # deepseek deepseek2
"model.layers.{bid}.feed_forward.shared_expert.gate_proj", # llama4
"model.layers.{bid}.mlp.shared_mlp.gate_proj", # hunyuan
),
# Feed-forward down
@ -447,11 +450,13 @@ class TensorNameMap:
"model.layers.{bid}.mlp.shared_experts.down_proj", # deepseek deepseek2
"model.layers.{bid}.feed_forward.shared_expert.down_proj", # llama4
"model.layers.{bid}.shared_mlp.output_linear", # granitemoe
"model.layers.{bid}.mlp.shared_mlp.down_proj", # hunyuan
),
MODEL_TENSOR.ATTN_Q_NORM: (
"language_model.encoder.layers.{bid}.self_attention.q_layernorm",
"model.layers.{bid}.self_attn.q_layernorm", # persimmon
"model.layers.{bid}.self_attn.query_layernorm", # hunyuan
"model.layers.{bid}.self_attn.q_norm", # cohere olmoe chameleon olmo2
"transformer.blocks.{bid}.attn.q_ln", # sea-lion
"encoder.layer.{bid}.attention.self.layer_norm_q", # jina-bert-v2
@ -461,6 +466,7 @@ class TensorNameMap:
MODEL_TENSOR.ATTN_K_NORM: (
"language_model.encoder.layers.{bid}.self_attention.k_layernorm",
"model.layers.{bid}.self_attn.k_layernorm", # persimmon
"model.layers.{bid}.self_attn.key_layernorm", # hunyuan
"model.layers.{bid}.self_attn.k_norm", # cohere olmoe chameleon olmo2
"transformer.blocks.{bid}.attn.k_ln", # sea-lion
"encoder.layer.{bid}.attention.self.layer_norm_k", # jina-bert-v2