ayo/ipex-llm
1
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 1

Modify arc nightly perf to fp16 (#11275)

Browse source 
* change api

* move to pr mode and remove the build

* add batch4 yaml and remove the bigcode

* remove batch4

* revert the starcode

* remove the exclude

* revert

---------

Co-authored-by: Yishuo Wang <yishuo.wang@intel.com>
This commit is contained in:
Wenjing Margaret Mao 2024-06-17 13:47:22 +08:00 committed by GitHub
parent a2a5890b48
commit bca5cbd96c
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
4 changed files with 4 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
python/llm/test/benchmark/arc-perf-test-batch2.yaml
View file

@ -29,7 +29,7 @@ in_out_pairs:
- '1024-128'
- '2048-256'
test_api:
- "transformer_int4_gpu" # on Intel GPU
- "transformer_int4_fp16_gpu" # on Intel GPU
cpu_embedding: False # whether put embedding to CPU (only avaiable now for gpu win related test_api)
exclude:
- 'bigcode/starcoder-15.5b-4bit:2048'

2
python/llm/test/benchmark/arc-perf-test.yaml
View file

@ -29,7 +29,7 @@ in_out_pairs:
- '1024-128'
- '2048-256'
test_api:
- "transformer_int4_gpu" # on Intel GPU
- "transformer_int4_fp16_gpu" # on Intel GPU
cpu_embedding: False # whether put embedding to CPU (only avaiable now for gpu win related test_api)
exclude:
# - 'fnlp/moss-moon-003-sft-4bit:1024'

2
python/llm/test/benchmark/arc-perf-transformers-437-batch2.yaml
View file

@ -15,5 +15,5 @@ in_out_pairs:
- '1024-128'
- '2048-256'
test_api:
- "transformer_int4_gpu" # on Intel GPU
- "transformer_int4_fp16_gpu" # on Intel GPU
cpu_embedding: False # whether put embedding to CPU (only avaiable now for gpu win related test_api)

2
python/llm/test/benchmark/arc-perf-transformers-437.yaml
View file

@ -15,5 +15,5 @@ in_out_pairs:
- '1024-128'
- '2048-256'
test_api:
- "transformer_int4_gpu" # on Intel GPU
- "transformer_int4_fp16_gpu" # on Intel GPU
cpu_embedding: False # whether put embedding to CPU (only avaiable now for gpu win related test_api)