tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-07-28 13:20:27 -04:00
Code Issues Packages Projects Releases Wiki Activity

model: add Ernie 4.5 MoE support (#14658)

Browse Source 
* Add Ernie4.5 MoE

* Fix Flake errors.

* Properly encode/decode MoE layer step

* Correct tensor mappings (.weight)

* Pass and read n_ff_exp

* n_ff_shexp calculation and further minor changes

* Rope fixes.

* .gitignore fix

* Add unit32 cast for Linux builds

* Apply suggestions from code review

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>

* Further fixes from code review

* Fix trailing whitespace

* Reenable missing experts error

* Code style from code review

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>

* Fix non-MoE regression

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>

---------

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>
This commit is contained in:
Piotr Wilkin (ilintar)
2025-07-17 23:15:32 +02:00
committed by GitHub
parent d6fb3f6b49
commit cb887f1bc1
7 changed files with 373 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
gguf-py/gguf/constants.py
View File

@@ -364,6 +364,7 @@ class MODEL_ARCH(IntEnum):
DOTS1 = auto()
ARCEE = auto()
ERNIE4_5 = auto()
ERNIE4_5_MOE = auto()
HUNYUAN_MOE = auto()
SMOLLM3 = auto()
LFM2 = auto()
@@ -680,6 +681,7 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
MODEL_ARCH.DOTS1: "dots1",
MODEL_ARCH.ARCEE: "arcee",
MODEL_ARCH.ERNIE4_5: "ernie4_5",
MODEL_ARCH.ERNIE4_5_MOE: "ernie4_5-moe",
MODEL_ARCH.FALCON_H1: "falcon-h1",
MODEL_ARCH.HUNYUAN_MOE: "hunyuan-moe",
MODEL_ARCH.SMOLLM3: "smollm3",
@@ -2022,6 +2024,28 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
MODEL_TENSOR.FFN_UP_SHEXP,
MODEL_TENSOR.FFN_EXP_PROBS_B,
],
MODEL_ARCH.ERNIE4_5_MOE: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.OUTPUT_NORM,
MODEL_TENSOR.OUTPUT,
MODEL_TENSOR.ATTN_NORM,
MODEL_TENSOR.ATTN_Q,
MODEL_TENSOR.ATTN_K,
MODEL_TENSOR.ATTN_V,
MODEL_TENSOR.ATTN_OUT,
MODEL_TENSOR.FFN_NORM,
MODEL_TENSOR.FFN_GATE,
MODEL_TENSOR.FFN_DOWN,
MODEL_TENSOR.FFN_UP,
MODEL_TENSOR.FFN_GATE_INP,
MODEL_TENSOR.FFN_GATE_EXP,
MODEL_TENSOR.FFN_DOWN_EXP,
MODEL_TENSOR.FFN_UP_EXP,
MODEL_TENSOR.FFN_GATE_SHEXP,
MODEL_TENSOR.FFN_DOWN_SHEXP,
MODEL_TENSOR.FFN_UP_SHEXP,
MODEL_TENSOR.FFN_EXP_PROBS_B,
],
MODEL_ARCH.PLM: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.OUTPUT,