huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-09-12 04:44:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix: impove adapter merge comments and remove unused conditional

Browse Source
This commit is contained in:
drbh 2024-07-18 18:20:51 +00:00
parent d27131bfa8
commit 59022c22b4
1 changed files with 7 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
server/text_generation_server/models/__init__.py
View File

@ -1153,11 +1153,15 @@ def get_model(
target_to_layer = build_layer_weight_lookup(model.model)
for index, adapter in enumerate(lora_adapters):
# currenly we only load one adapter at a time but
# this can be extended to merge multiple adapters
# The AdapterParameters object allows for merging multiple adapters into a single adapter.
# At the moment, we only support loading a single adapter into the model, but we keep the
# AdapterParameters object for easier extension in the future.
adapter_parameters = AdapterParameters(
adapter_info=[adapter],
weights=None, # will be set to 1
# when merging multiple adapters we can weight them differently
# if this is not set, all adapters will be weighted equally
# see: text_generation_server.utils.merges.strategies for impl
weights=None,
merge_strategy=0,
density=1.0,
majority_sign_method=0,
@ -1166,9 +1170,6 @@ def get_model(
adapter_index = index + 1
adapter_to_index[adapter.id] = adapter_index
if adapter_index in model.loaded_adapters:
continue
logger.info(
f"Loading adapter weights into model: {','.join([adapter.id for adapter in adapter_parameters.adapter_info])}"
)