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

Merge branch 'master' of https://github.com/namtranase/llama.cpp

Browse source
This commit is contained in:
namtranase 2023-12-25 17:13:50 +07:00
commit 44f4ce2272
3 changed files with 3 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

2
awq-py/README.md
View file

@ -117,4 +117,4 @@ We use three types of llamacpp quantization methods to work with our version, in
|MPT 7B | bits/weight | 16.0 | 4.5 | 5.0 | 2.6 |
|AWQ-MPT 7B| perplexity | 8.4944 | 8.7053 | 8.6750 | 10.2873|
|AWQ-MPT 7B| file size | 13.7G | 3.9G | 4.3G | 2.8G |
|AWQ-MPT 7B| bits/weight | 16.0 | 4.5 | 5.0 | 2.6 |
|AWQ-MPT 7B| bits/weight | 16.0 | 4.5 | 5.0 | 2.6 |

5
awq-py/awq/apply_awq.py
View file

@ -189,10 +189,7 @@ def apply_scale(module, scales_list, input_feat_dict=None):
if isinstance(prev_op, nn.Linear):
assert len(layers) == 1
scale_fc_fc(prev_op, layers[0], scales)
elif (
isinstance(prev_op, (nn.LayerNorm, LlamaRMSNorm))
or "rmsnorm" in str(prev_op.__class__).lower()
):
elif isinstance(prev_op, (nn.LayerNorm, LlamaRMSNorm)) or "rmsnorm" in str(prev_op.__class__).lower():
scale_ln_fcs(prev_op, layers, scales)
elif isinstance(prev_op, (nn.GELU, BloomGelu, GELUActivation)):
new_module = ScaledActivation(prev_op, scales)

2
convert-hf-to-gguf.py
View file

@ -1041,7 +1041,7 @@ dir_model = args.model
if args.awq_path:
sys.path.insert(1, str(Path(__file__).parent / 'awq-py'))
from awq.apply_awq import add_scale_weights
from awq.apply_awq import add_scale_weights
tmp_model_path = args.model / "weighted_model"
dir_model = tmp_model_path
if tmp_model_path.is_dir():