tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-08-20 06:36:48 -04:00
Code Issues Packages Projects Releases Wiki Activity

llama : add support for the cohere2 model architecture (#10900)

Browse Source
This commit is contained in:
DAN™
2025-01-04 09:33:31 -05:00
committed by GitHub
parent 78c6785175
commit 46be942214
6 changed files with 221 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
gguf-py/gguf/constants.py
View File

@@ -255,6 +255,7 @@ class MODEL_ARCH(IntEnum):
MAMBA = auto()
XVERSE = auto()
COMMAND_R = auto()
COHERE2 = auto()
DBRX = auto()
OLMO = auto()
OLMO2 = auto()
@@ -437,6 +438,7 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
MODEL_ARCH.MAMBA: "mamba",
MODEL_ARCH.XVERSE: "xverse",
MODEL_ARCH.COMMAND_R: "command-r",
MODEL_ARCH.COHERE2: "cohere2",
MODEL_ARCH.DBRX: "dbrx",
MODEL_ARCH.OLMO: "olmo",
MODEL_ARCH.OLMO2: "olmo2",
@@ -1136,6 +1138,18 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
MODEL_TENSOR.ATTN_K_NORM,
MODEL_TENSOR.ATTN_Q_NORM,
],
MODEL_ARCH.COHERE2: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.OUTPUT_NORM,
MODEL_TENSOR.ATTN_NORM,
MODEL_TENSOR.ATTN_Q,
MODEL_TENSOR.ATTN_K,
MODEL_TENSOR.ATTN_V,
MODEL_TENSOR.ATTN_OUT,
MODEL_TENSOR.FFN_GATE,
MODEL_TENSOR.FFN_DOWN,
MODEL_TENSOR.FFN_UP,
],
MODEL_ARCH.DBRX: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.OUTPUT_NORM,