ipex-llm/python/llm/test/benchmark/arc-perf-test-batch4.yaml
Wenjing Margaret Mao b2f62a8561
Add batch 4 perf test (#11355)
* copy files to this branch

* add tasks

* comment one model

* change the model to test the 4.36

* only test batch-4

* typo

* typo

* typo

* typo

* typo

* typo

* add 4.37-batch4

* change the file name

* revet yaml file

* no print

* add batch4 task

* revert

---------

Co-authored-by: Yishuo Wang <yishuo.wang@intel.com>
2024-06-20 09:48:52 +08:00

55 lines
2 KiB
YAML

repo_id:
- 'meta-llama/Llama-2-7b-chat-hf'
- 'meta-llama/Llama-2-13b-chat-hf'
- 'THUDM/chatglm2-6b'
- 'THUDM/chatglm3-6b-4bit'
- 'tiiuae/falcon-7b-instruct-with-patch'
- 'mosaicml/mpt-7b-chat'
- 'redpajama/gptneox-7b-redpajama-bf16'
- 'bigcode/starcoder-15.5b-4bit'
- 'databricks/dolly-v1-6b'
- 'databricks/dolly-v2-7b'
- 'databricks/dolly-v2-12b'
- 'internlm/internlm-chat-7b'
- 'Qwen/Qwen-7B-Chat'
- 'BAAI/AquilaChat-7B'
- 'baichuan-inc/Baichuan2-7B-Chat'
- 'baichuan-inc/Baichuan2-13B-Chat-4bit'
- 'bigscience/bloomz-7b1'
# - 'fnlp/moss-moon-003-sft-4bit' # moss-moon-003-sft cannot work on transformers 4.34+
- 'mistralai/Mistral-7B-v0.1' #mwj: need to check
local_model_hub: '/mnt/disk1/models'
warm_up: 1
num_trials: 3
num_beams: 1 # default to greedy search
low_bit: 'sym_int4' # default to use 'sym_int4' (i.e. symmetric int4)
batch_size: 4 # default to 1
in_out_pairs:
- '32-32'
- '1024-128'
- '2048-256'
test_api:
- "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:
- 'meta-llama/Llama-2-13b-chat-hf:2048'
- 'tiiuae/falcon-7b-instruct-with-patch:2048'
- 'mosaicml/mpt-7b-chat:2048'
- 'redpajama/gptneox-7b-redpajama-bf16:2048'
- 'bigcode/starcoder-15.5b-4bit:1024'
- 'bigcode/starcoder-15.5b-4bit:2048'
- 'databricks/dolly-v1-6b:2048'
- 'databricks/dolly-v2-7b:2048'
- 'databricks/dolly-v2-12b:1024'
- 'databricks/dolly-v2-12b:2048'
- 'internlm/internlm-chat-7b:2048'
- 'Qwen/Qwen-7B-Chat:2048'
- 'BAAI/AquilaChat-7B:2048'
- 'baichuan-inc/Baichuan2-7B-Chat:2048'
- 'baichuan-inc/Baichuan2-13B-Chat-4bit:1024'
- 'baichuan-inc/Baichuan2-13B-Chat-4bit:2048'
- 'bigscience/bloomz-7b1:1024'
- 'bigscience/bloomz-7b1:2048'
# - 'fnlp/moss-moon-003-sft-4bit:1024'
# - 'fnlp/moss-moon-003-sft-4bit:2048'
task: 'continuation' # task can be 'continuation', 'QA' and 'summarize'