mirror of
https://github.com/kvcache-ai/ktransformers.git
synced 2026-04-20 06:18:59 +00:00
42 lines
931 B
YAML
42 lines
931 B
YAML
global_vars:
|
|
q8: &q8 /home/dataset/int8
|
|
q4: &q4 /home/dataset/int4
|
|
w8a8: &w8a8 /home/dataset/w8a8
|
|
ascend_toolkit: &toolkit /usr/local/Ascend/ascend-toolkit/set_env.sh
|
|
atb: &atb /home/ascend/ascend-transformer-boost/output/atb/set_env.sh
|
|
optimize_config_path: &opt_cfg ./ktransformers/optimize/optimize_rules/DeepSeek-V3-Chat-800IA2-npu.yaml
|
|
|
|
hyperparams:
|
|
- cpu_infer: 49
|
|
w8a8_safetensor_path: *w8a8
|
|
|
|
- cpu_infer: 37
|
|
|
|
export_area:
|
|
ASDOPS_LOG_TO_FILE: 0
|
|
ASDOPS_LOG_TO_STDOUT: 0
|
|
ASDOPS_LOG_LEVEL: ERROR
|
|
ATB_LOG_TO_FILE: 0
|
|
ATB_LOG_TO_STDOUT: 0
|
|
ATB_LOG_LEVEL: ERROR
|
|
USE_MERGE: 0
|
|
|
|
source_area:
|
|
- *toolkit
|
|
- *atb
|
|
|
|
torchrun_area:
|
|
long:
|
|
nproc_per_node: 1
|
|
short:
|
|
m: ktransformers.local_chat
|
|
|
|
model_area:
|
|
cpu_infer: $cpu_infer
|
|
model_path: $model_path
|
|
gguf_path: *q8
|
|
q4_gguf_path: *q4
|
|
w8a8_safetensor_path: $w8a8_safetensor_path
|
|
max_new_tokens: 50
|
|
optimize_config_path: *opt_cfg
|