tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-07-30 14:13:57 -04:00
Code Issues Packages Projects Releases Wiki Activity

stablelm : StableLM support (#3586)

Browse Source 
* Add support for stablelm-3b-4e1t
* Supports GPU offloading of (n-1) layers
This commit is contained in:
Galunid
2023-11-14 11:17:12 +01:00
committed by GitHub
parent b46d12f86d
commit 36eed0c42c
6 changed files with 322 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
gguf-py/gguf/constants.py
View File

@@ -90,6 +90,7 @@ class MODEL_ARCH(IntEnum):
REFACT = auto()
BERT = auto()
BLOOM = auto()
STABLELM = auto()
class MODEL_TENSOR(IntEnum):
@@ -129,6 +130,7 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
MODEL_ARCH.REFACT: "refact",
MODEL_ARCH.BERT: "bert",
MODEL_ARCH.BLOOM: "bloom",
MODEL_ARCH.STABLELM: "stablelm",
}
TENSOR_NAMES: dict[MODEL_TENSOR, str] = {
@@ -299,6 +301,21 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
MODEL_TENSOR.FFN_DOWN,
MODEL_TENSOR.FFN_UP,
],
MODEL_ARCH.STABLELM: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.OUTPUT_NORM,
MODEL_TENSOR.OUTPUT,
MODEL_TENSOR.ROPE_FREQS,
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.GPT2: [
# TODO
],