evalscope 0.5.0__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- evalscope/__init__.py +3 -0
- evalscope/backend/__init__.py +3 -0
- evalscope/backend/base.py +27 -0
- evalscope/backend/opencompass/__init__.py +3 -0
- evalscope/backend/opencompass/api_meta_template.py +64 -0
- evalscope/backend/opencompass/backend_manager.py +247 -0
- evalscope/backend/opencompass/tasks/__init__.py +1 -0
- evalscope/backend/opencompass/tasks/eval_api.py +30 -0
- evalscope/backend/opencompass/tasks/eval_datasets.py +71 -0
- evalscope/backend/vlm_eval_kit/__init__.py +1 -0
- evalscope/backend/vlm_eval_kit/backend_manager.py +153 -0
- evalscope/benchmarks/__init__.py +4 -0
- evalscope/benchmarks/arc/__init__.py +5 -0
- evalscope/benchmarks/arc/ai2_arc.py +148 -0
- evalscope/benchmarks/arc/arc_adapter.py +231 -0
- evalscope/benchmarks/bbh/__init__.py +6 -0
- evalscope/benchmarks/bbh/bbh_adapter.py +308 -0
- evalscope/benchmarks/bbh/cot_prompts/boolean_expressions.txt +23 -0
- evalscope/benchmarks/bbh/cot_prompts/causal_judgement.txt +25 -0
- evalscope/benchmarks/bbh/cot_prompts/date_understanding.txt +33 -0
- evalscope/benchmarks/bbh/cot_prompts/disambiguation_qa.txt +37 -0
- evalscope/benchmarks/bbh/cot_prompts/dyck_languages.txt +72 -0
- evalscope/benchmarks/bbh/cot_prompts/formal_fallacies.txt +44 -0
- evalscope/benchmarks/bbh/cot_prompts/geometric_shapes.txt +78 -0
- evalscope/benchmarks/bbh/cot_prompts/hyperbaton.txt +28 -0
- evalscope/benchmarks/bbh/cot_prompts/logical_deduction_five_objects.txt +37 -0
- evalscope/benchmarks/bbh/cot_prompts/logical_deduction_seven_objects.txt +37 -0
- evalscope/benchmarks/bbh/cot_prompts/logical_deduction_three_objects.txt +37 -0
- evalscope/benchmarks/bbh/cot_prompts/movie_recommendation.txt +42 -0
- evalscope/benchmarks/bbh/cot_prompts/multistep_arithmetic_two.txt +25 -0
- evalscope/benchmarks/bbh/cot_prompts/navigate.txt +43 -0
- evalscope/benchmarks/bbh/cot_prompts/object_counting.txt +37 -0
- evalscope/benchmarks/bbh/cot_prompts/penguins_in_a_table.txt +41 -0
- evalscope/benchmarks/bbh/cot_prompts/reasoning_about_colored_objects.txt +63 -0
- evalscope/benchmarks/bbh/cot_prompts/ruin_names.txt +44 -0
- evalscope/benchmarks/bbh/cot_prompts/salient_translation_error_detection.txt +40 -0
- evalscope/benchmarks/bbh/cot_prompts/snarks.txt +30 -0
- evalscope/benchmarks/bbh/cot_prompts/sports_understanding.txt +10 -0
- evalscope/benchmarks/bbh/cot_prompts/temporal_sequences.txt +77 -0
- evalscope/benchmarks/bbh/cot_prompts/tracking_shuffled_objects_five_objects.txt +40 -0
- evalscope/benchmarks/bbh/cot_prompts/tracking_shuffled_objects_seven_objects.txt +40 -0
- evalscope/benchmarks/bbh/cot_prompts/tracking_shuffled_objects_three_objects.txt +40 -0
- evalscope/benchmarks/bbh/cot_prompts/web_of_lies.txt +28 -0
- evalscope/benchmarks/bbh/cot_prompts/word_sorting.txt +17 -0
- evalscope/benchmarks/benchmark.py +65 -0
- evalscope/benchmarks/ceval/__init__.py +5 -0
- evalscope/benchmarks/ceval/ceval_adapter.py +340 -0
- evalscope/benchmarks/ceval/ceval_exam.py +159 -0
- evalscope/benchmarks/cmmlu/__init__.py +5 -0
- evalscope/benchmarks/cmmlu/cmmlu.py +166 -0
- evalscope/benchmarks/cmmlu/cmmlu_adapter.py +369 -0
- evalscope/benchmarks/competition_math/__init__.py +5 -0
- evalscope/benchmarks/competition_math/competition_math.py +88 -0
- evalscope/benchmarks/competition_math/competition_math_adapter.py +470 -0
- evalscope/benchmarks/data_adapter.py +263 -0
- evalscope/benchmarks/general_qa/__init__.py +5 -0
- evalscope/benchmarks/general_qa/general_qa_adapter.py +186 -0
- evalscope/benchmarks/gsm8k/__init__.py +5 -0
- evalscope/benchmarks/gsm8k/gsm8k.py +127 -0
- evalscope/benchmarks/gsm8k/gsm8k_adapter.py +236 -0
- evalscope/benchmarks/hellaswag/__init__.py +5 -0
- evalscope/benchmarks/hellaswag/hellaswag.py +116 -0
- evalscope/benchmarks/hellaswag/hellaswag_adapter.py +222 -0
- evalscope/benchmarks/humaneval/__init__.py +5 -0
- evalscope/benchmarks/humaneval/humaneval.py +82 -0
- evalscope/benchmarks/humaneval/humaneval_adapter.py +21 -0
- evalscope/benchmarks/mmlu/__init__.py +5 -0
- evalscope/benchmarks/mmlu/mmlu.py +174 -0
- evalscope/benchmarks/mmlu/mmlu_adapter.py +375 -0
- evalscope/benchmarks/race/__init__.py +5 -0
- evalscope/benchmarks/race/race.py +118 -0
- evalscope/benchmarks/race/race_adapter.py +229 -0
- evalscope/benchmarks/trivia_qa/__init__.py +5 -0
- evalscope/benchmarks/trivia_qa/trivia_qa.py +104 -0
- evalscope/benchmarks/trivia_qa/trivia_qa_adapter.py +207 -0
- evalscope/benchmarks/truthful_qa/__init__.py +5 -0
- evalscope/benchmarks/truthful_qa/truthful_qa.py +167 -0
- evalscope/benchmarks/truthful_qa/truthful_qa_adapter.py +351 -0
- evalscope/cache.py +98 -0
- evalscope/cli/__init__.py +1 -0
- evalscope/cli/base.py +20 -0
- evalscope/cli/cli.py +26 -0
- evalscope/cli/start_perf.py +37 -0
- evalscope/cli/start_server.py +138 -0
- evalscope/config.py +165 -0
- evalscope/constants.py +150 -0
- evalscope/evaluator/__init__.py +3 -0
- evalscope/evaluator/evaluator.py +689 -0
- evalscope/evaluator/rating_eval.py +178 -0
- evalscope/evaluator/reviewer/__init__.py +1 -0
- evalscope/evaluator/reviewer/auto_reviewer.py +411 -0
- evalscope/metrics/__init__.py +1 -0
- evalscope/metrics/bundled_rouge_score/__init__.py +14 -0
- evalscope/metrics/bundled_rouge_score/rouge_scorer.py +342 -0
- evalscope/metrics/code_metric.py +104 -0
- evalscope/metrics/math_accuracy.py +60 -0
- evalscope/metrics/metrics.py +405 -0
- evalscope/metrics/rouge_metric.py +129 -0
- evalscope/models/__init__.py +4 -0
- evalscope/models/custom/__init__.py +4 -0
- evalscope/models/custom/custom_model.py +53 -0
- evalscope/models/dummy_chat_model.py +50 -0
- evalscope/models/model.py +88 -0
- evalscope/models/model_adapter.py +586 -0
- evalscope/models/openai_model.py +103 -0
- evalscope/models/template.py +1446 -0
- evalscope/perf/__init__.py +0 -0
- evalscope/perf/_logging.py +32 -0
- evalscope/perf/api_plugin_base.py +60 -0
- evalscope/perf/custom_api.py +87 -0
- evalscope/perf/dashscope_api.py +84 -0
- evalscope/perf/dataset_plugin_base.py +64 -0
- evalscope/perf/datasets/__init__.py +0 -0
- evalscope/perf/datasets/line_by_line.py +18 -0
- evalscope/perf/datasets/longalpaca_12k.py +20 -0
- evalscope/perf/datasets/openqa.py +22 -0
- evalscope/perf/how_to_analysis_result.py +24 -0
- evalscope/perf/http_client.py +756 -0
- evalscope/perf/openai_api.py +130 -0
- evalscope/perf/plugin_registry.py +35 -0
- evalscope/perf/query_parameters.py +42 -0
- evalscope/perf/server_sent_event.py +43 -0
- evalscope/preprocess/__init__.py +1 -0
- evalscope/preprocess/tokenizers/__init__.py +0 -0
- evalscope/preprocess/tokenizers/gpt2_tokenizer.py +221 -0
- evalscope/registry/__init__.py +1 -0
- evalscope/registry/tasks/arc.yaml +29 -0
- evalscope/registry/tasks/bbh.yaml +27 -0
- evalscope/registry/tasks/bbh_mini.yaml +27 -0
- evalscope/registry/tasks/ceval.yaml +27 -0
- evalscope/registry/tasks/ceval_mini.yaml +27 -0
- evalscope/registry/tasks/cmmlu.yaml +27 -0
- evalscope/registry/tasks/eval_qwen-7b-chat_v100.yaml +28 -0
- evalscope/registry/tasks/general_qa.yaml +27 -0
- evalscope/registry/tasks/gsm8k.yaml +29 -0
- evalscope/registry/tasks/mmlu.yaml +29 -0
- evalscope/registry/tasks/mmlu_mini.yaml +27 -0
- evalscope/run.py +404 -0
- evalscope/run_arena.py +204 -0
- evalscope/run_ms.py +140 -0
- evalscope/summarizer.py +144 -0
- evalscope/third_party/__init__.py +1 -0
- evalscope/third_party/toolbench_static/__init__.py +3 -0
- evalscope/third_party/toolbench_static/eval.py +219 -0
- evalscope/third_party/toolbench_static/infer.py +278 -0
- evalscope/third_party/toolbench_static/llm/__init__.py +1 -0
- evalscope/third_party/toolbench_static/llm/swift_infer.py +45 -0
- evalscope/third_party/toolbench_static/toolbench_static.py +50 -0
- evalscope/tools/__init__.py +1 -0
- evalscope/tools/combine_reports.py +140 -0
- evalscope/tools/gen_mmlu_subject_mapping.py +90 -0
- evalscope/tools/rewrite_eval_results.py +95 -0
- evalscope/utils/__init__.py +4 -0
- evalscope/utils/arena_utils.py +247 -0
- evalscope/utils/completion_parsers.py +87 -0
- evalscope/utils/logger.py +64 -0
- evalscope/utils/task_cfg_parser.py +10 -0
- evalscope/utils/task_utils.py +19 -0
- evalscope/utils/utils.py +625 -0
- evalscope/version.py +4 -0
- evalscope-0.5.0.dist-info/METADATA +566 -0
- evalscope-0.5.0.dist-info/RECORD +165 -0
- evalscope-0.5.0.dist-info/WHEEL +5 -0
- evalscope-0.5.0.dist-info/entry_points.txt +3 -0
- evalscope-0.5.0.dist-info/top_level.txt +1 -0
|
@@ -0,0 +1,64 @@
|
|
|
1
|
+
# Copyright (c) Alibaba, Inc. and its affiliates.
|
|
2
|
+
|
|
3
|
+
import logging
|
|
4
|
+
from typing import Optional
|
|
5
|
+
|
|
6
|
+
init_loggers = {}
|
|
7
|
+
|
|
8
|
+
formatter = logging.Formatter(
|
|
9
|
+
'%(asctime)s - %(name)s - %(levelname)s - %(message)s')
|
|
10
|
+
|
|
11
|
+
|
|
12
|
+
def get_logger(log_file: Optional[str] = None,
|
|
13
|
+
log_level: int = logging.INFO,
|
|
14
|
+
file_mode: str = 'w'):
|
|
15
|
+
""" Get logging logger
|
|
16
|
+
|
|
17
|
+
Args:
|
|
18
|
+
log_file: Log filename, if specified, file handler will be added to
|
|
19
|
+
logger
|
|
20
|
+
log_level: Logging level.
|
|
21
|
+
file_mode: Specifies the mode to open the file, if filename is
|
|
22
|
+
specified (if filemode is unspecified, it defaults to 'w').
|
|
23
|
+
"""
|
|
24
|
+
|
|
25
|
+
logger_name = __name__.split('.')[0]
|
|
26
|
+
logger = logging.getLogger(logger_name)
|
|
27
|
+
|
|
28
|
+
if logger_name in init_loggers:
|
|
29
|
+
add_file_handler_if_needed(logger, log_file, file_mode, log_level)
|
|
30
|
+
return logger
|
|
31
|
+
|
|
32
|
+
for handler in logger.root.handlers:
|
|
33
|
+
if type(handler) is logging.StreamHandler:
|
|
34
|
+
handler.setLevel(logging.ERROR)
|
|
35
|
+
|
|
36
|
+
stream_handler = logging.StreamHandler()
|
|
37
|
+
handlers = [stream_handler]
|
|
38
|
+
|
|
39
|
+
if log_file is not None:
|
|
40
|
+
file_handler = logging.FileHandler(log_file, file_mode)
|
|
41
|
+
handlers.append(file_handler)
|
|
42
|
+
|
|
43
|
+
for handler in handlers:
|
|
44
|
+
handler.setFormatter(formatter)
|
|
45
|
+
handler.setLevel(log_level)
|
|
46
|
+
logger.addHandler(handler)
|
|
47
|
+
|
|
48
|
+
logger.setLevel(log_level)
|
|
49
|
+
|
|
50
|
+
init_loggers[logger_name] = True
|
|
51
|
+
|
|
52
|
+
return logger
|
|
53
|
+
|
|
54
|
+
|
|
55
|
+
def add_file_handler_if_needed(logger, log_file, file_mode, log_level):
|
|
56
|
+
for handler in logger.handlers:
|
|
57
|
+
if isinstance(handler, logging.FileHandler):
|
|
58
|
+
return
|
|
59
|
+
|
|
60
|
+
if log_file is not None:
|
|
61
|
+
file_handler = logging.FileHandler(log_file, file_mode)
|
|
62
|
+
file_handler.setFormatter(formatter)
|
|
63
|
+
file_handler.setLevel(log_level)
|
|
64
|
+
logger.addHandler(file_handler)
|
|
@@ -0,0 +1,19 @@
|
|
|
1
|
+
# Copyright (c) Alibaba, Inc. and its affiliates.
|
|
2
|
+
from enum import Enum
|
|
3
|
+
|
|
4
|
+
|
|
5
|
+
class EvalBackend(Enum):
|
|
6
|
+
# Use native evaluation pipeline of Eval-Scope
|
|
7
|
+
NATIVE = 'Native'
|
|
8
|
+
|
|
9
|
+
# Use OpenCompass framework as the evaluation backend
|
|
10
|
+
OPEN_COMPASS = 'OpenCompass'
|
|
11
|
+
|
|
12
|
+
# Use VLM Eval Kit as the multi-modal model evaluation backend
|
|
13
|
+
VLM_EVAL_KIT = 'VLMEvalKit'
|
|
14
|
+
|
|
15
|
+
# Use third-party evaluation backend/modules
|
|
16
|
+
THIRD_PARTY = 'ThirdParty'
|
|
17
|
+
|
|
18
|
+
|
|
19
|
+
|