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

gemma2: add sliding window mask (#8227)

Browse source 
* gemma2: add sliding window mask

* fix data_swa uninitialized

* better naming

* add co-author

Co-authored-by: Arlo Phoenix <arlo-phoenix@users.noreply.github.com>

* replace list with single tensor

* update

* llama : minor styling

* convert : add sanity check for query_pre_attn_scalar

* fix small typo in README

---------

Co-authored-by: Arlo Phoenix <arlo-phoenix@users.noreply.github.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
This commit is contained in:
Xuan Son Nguyen 2024-07-01 18:48:34 +02:00 • committed by GitHub
parent 0ddeff1023
commit 49122a873f
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
5 changed files with 79 additions and 32 deletions
Download patch file Download diff file Expand all files Collapse all files

6
convert-hf-to-gguf.py
View file

@ -2369,6 +2369,12 @@ class Gemma2Model(Model):
self.gguf_writer.add_final_logit_softcapping(
self.hparams["final_logit_softcapping"]
)
self.gguf_writer.add_sliding_window(self.hparams["sliding_window"])
# sanity check
attn_scalar = self.hparams["query_pre_attn_scalar"]
if attn_scalar != hparams["hidden_size"] / hparams["num_attention_heads"]:
raise ValueError("query_pre_attn_scalar must be equal to n_embd / n_head")
def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iterable[tuple[str, Tensor]]:
del bid # unusem