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

edit 'ppl_result does not exist' issue, delete useless code (#10767)

Browse source 
* edit ppl_result not exist issue, delete useless code

* delete nonzero_min function

---------

Co-authored-by: jenniew <jenniewang123@gmail.com>
This commit is contained in:
Wenjing Margaret Mao 2024-04-16 18:11:56 +08:00 committed by GitHub
parent 899d392e2f
commit c41730e024
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 0 additions and 26 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

26
python/llm/test/benchmark/perplexity/ppl_csv_to_html.py
View file

@ -33,26 +33,9 @@ def highlight_vals(val, max=3.0, color1='red', color2='green', color3='yellow',
else: else:
return '' return ''
def nonzero_min(lst):
non_zero_lst = [num for num in lst if num > 0.0]
return min(non_zero_lst) if non_zero_lst else None
def is_diffs_within_normal_range(diff_ppl_result, threshold=5.0): def is_diffs_within_normal_range(diff_ppl_result, threshold=5.0):
return not any(diff < (-threshold) for diff in diff_ppl_result if isinstance(diff, float)) return not any(diff < (-threshold) for diff in diff_ppl_result if isinstance(diff, float))
def add_to_dict(dict, key, value):
if key not in dict:
dict[key] = []
dict[key].append(value)
def best_in_dict(dict, key, value):
if key in dict:
best_value = nonzero_min(dict[key])
if best_value < value or value <= 0.0:
return best_value
return value
return value
def create_fp16_dict(fp16_path): def create_fp16_dict(fp16_path):
fp16_df = pd.read_csv(fp16_path) fp16_df = pd.read_csv(fp16_path)
fp16_dict = {} fp16_dict = {}
@ -118,15 +101,6 @@ def main():
diff_ppl_result=['']*len(latest_csv.index) diff_ppl_result=['']*len(latest_csv.index)
ppl_result = 'ppl_result' ppl_result = 'ppl_result'
csv_dict = {}
for csv_file in csv_files:
current_csv = pd.read_csv(csv_file, index_col=0)
for current_csv_ind,current_csv_row in current_csv.iterrows():
current_csv_model=current_csv_row['Model'].strip()
current_csv_precision=current_csv_row['Precision'].strip()
current_csv_model_ppl_result=current_csv_model+'-'+current_csv_precision+'-'+'ppl_result'
add_to_dict(csv_dict, current_csv_model_ppl_result, current_csv_row[ppl_result])
for latest_csv_ind,latest_csv_row in latest_csv.iterrows(): for latest_csv_ind,latest_csv_row in latest_csv.iterrows():