tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-08-12 03:21:10 -04:00
Code Issues Packages Projects Releases Wiki Activity

llama : support upcoming Qwen2 (#5037)

Browse Source
This commit is contained in:
Shijie
2024-01-19 19:53:13 +08:00
committed by GitHub
parent de9a147df1
commit 9b75cb2b3c
3 changed files with 211 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
gguf-py/gguf/constants.py
View File

@@ -97,6 +97,7 @@ class MODEL_ARCH(IntEnum):
BLOOM = auto()
STABLELM = auto()
QWEN = auto()
QWEN2 = auto()
PHI2 = auto()
PLAMO = auto()
CODESHELL = auto()
@@ -146,6 +147,7 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
MODEL_ARCH.BLOOM: "bloom",
MODEL_ARCH.STABLELM: "stablelm",
MODEL_ARCH.QWEN: "qwen",
MODEL_ARCH.QWEN2: "qwen2",
MODEL_ARCH.PHI2: "phi2",
MODEL_ARCH.PLAMO: "plamo",
MODEL_ARCH.CODESHELL: "codeshell",
@@ -358,6 +360,20 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
MODEL_TENSOR.FFN_DOWN,
MODEL_TENSOR.FFN_UP,
],
MODEL_ARCH.QWEN2: [
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_ARCH.PLAMO: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.OUTPUT_NORM,