vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

llama: Add attention and final logit soft-capping, update scaling factor to Gemma2 (#8197)

Browse source 
* Add attention and final logit softcapping.

* fix

* Add custom add_ functions

* Disable flash attention for Gemma2

* Update src/llama.cpp

Co-authored-by: slaren <slarengh@gmail.com>

* Add default value for attention and final logit softcap value

* Add custom kq scaling from Gemma2Attention

* Remove custom pre attention scaling and use computed value instead.

---------

Co-authored-by: slaren <slarengh@gmail.com>
This commit is contained in:
Andrei 2024-06-29 20:44:08 -07:00 • committed by GitHub
parent 72272b83a3
commit 1c5eba6f8e
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
4 changed files with 46 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

2
gguf-py/gguf/constants.py
View file

@ -50,6 +50,8 @@ class Keys:
POOLING_TYPE = "{arch}.pooling_type"
LOGIT_SCALE = "{arch}.logit_scale"
DECODER_START_TOKEN_ID = "{arch}.decoder_start_token_id"
ATTN_LOGIT_SOFTCAPPING = "{arch}.attn_logit_softcapping"
FINAL_LOGIT_SOFTCAPPING = "{arch}.final_logit_softcapping"
class Attention:
HEAD_COUNT = "{arch}.attention.head_count"

6
gguf-py/gguf/gguf_writer.py
View file

@ -516,6 +516,12 @@ class GGUFWriter:
def add_logit_scale(self, value: float) -> None:
self.add_float32(Keys.LLM.LOGIT_SCALE.format(arch=self.arch), value)
def add_attn_logit_softcapping(self, value: float) -> None:
self.add_float32(Keys.LLM.ATTN_LOGIT_SOFTCAPPING.format(arch=self.arch), value)
def add_final_logit_softcapping(self, value: float) -> None:
self.add_float32(Keys.LLM.FINAL_LOGIT_SOFTCAPPING.format(arch=self.arch), value)
def add_expert_count(self, count: int) -> None:
self.add_uint32(Keys.LLM.EXPERT_COUNT.format(arch=self.arch), count)