Spaces:
AIR-Bench
/
Running on CPU Upgrade

hanhainebula commited on
Commit
606d718
1 Parent(s): 77ded94

Modify the evaluation steps

Browse files
Files changed (1) hide show
  1. src/about.py +7 -14
src/about.py CHANGED
@@ -19,20 +19,15 @@ EVALUATION_QUEUE_TEXT = """
19
 
20
  1. Install AIR-Bench
21
  ```bash
22
- # Clone the repo
23
- git clone https://github.com/AIR-Bench/AIR-Bench.git
24
-
25
- # Install the package
26
- cd AIR-Bench
27
- pip install .
28
  ```
29
  2. Run the evaluation script
30
  ```bash
31
  cd AIR-Bench/scripts
32
  # Run all tasks
33
- python run_AIR-Bench.py \\
34
  --output_dir ./search_results \\
35
- --encoder BAAI/bge-m3 \
36
  --reranker BAAI/bge-reranker-v2-m3 \\
37
  --search_top_k 1000 \\
38
  --rerank_top_k 100 \\
@@ -46,7 +41,7 @@ python run_AIR-Bench.py \\
46
  --overwrite False
47
 
48
  # Run the tasks in the specified task type
49
- python run_AIR-Bench.py \\
50
  --task_types long-doc \\
51
  --output_dir ./search_results \\
52
  --encoder BAAI/bge-m3 \\
@@ -63,7 +58,7 @@ python run_AIR-Bench.py \\
63
  --overwrite False
64
 
65
  # Run the tasks in the specified task type and domains
66
- python run_AIR-Bench.py \\
67
  --task_types long-doc \\
68
  --domains arxiv book \\
69
  --output_dir ./search_results \\
@@ -81,7 +76,7 @@ python run_AIR-Bench.py \\
81
  --overwrite False
82
 
83
  # Run the tasks in the specified languages
84
- python run_AIR-Bench.py \\
85
  --languages en \\
86
  --output_dir ./search_results \\
87
  --encoder BAAI/bge-m3 \\
@@ -98,15 +93,13 @@ python run_AIR-Bench.py \\
98
  --overwrite False
99
 
100
  # Run the tasks in the specified task type, domains, and languages
101
- python run_AIR-Bench.py \\
102
  --task_types qa \\
103
  --domains wiki web \\
104
  --languages en \\
105
  --output_dir ./search_results \\
106
  --encoder BAAI/bge-m3 \\
107
- --encoder_link https://huggingface.co/BAAI/bge-m3 \\
108
  --reranker BAAI/bge-reranker-v2-m3 \\
109
- --reranker_link https://huggingface.co/BAAI/bge-reranker-v2-m3 \\
110
  --search_top_k 1000 \\
111
  --rerank_top_k 100 \\
112
  --max_query_length 512 \\
 
19
 
20
  1. Install AIR-Bench
21
  ```bash
22
+ pip install air-benchmark
 
 
 
 
 
23
  ```
24
  2. Run the evaluation script
25
  ```bash
26
  cd AIR-Bench/scripts
27
  # Run all tasks
28
+ python run_air_benchmark.py \\
29
  --output_dir ./search_results \\
30
+ --encoder BAAI/bge-m3 \\
31
  --reranker BAAI/bge-reranker-v2-m3 \\
32
  --search_top_k 1000 \\
33
  --rerank_top_k 100 \\
 
41
  --overwrite False
42
 
43
  # Run the tasks in the specified task type
44
+ python run_air_benchmark.py \\
45
  --task_types long-doc \\
46
  --output_dir ./search_results \\
47
  --encoder BAAI/bge-m3 \\
 
58
  --overwrite False
59
 
60
  # Run the tasks in the specified task type and domains
61
+ python run_air_benchmark.py \\
62
  --task_types long-doc \\
63
  --domains arxiv book \\
64
  --output_dir ./search_results \\
 
76
  --overwrite False
77
 
78
  # Run the tasks in the specified languages
79
+ python run_air_benchmark.py \\
80
  --languages en \\
81
  --output_dir ./search_results \\
82
  --encoder BAAI/bge-m3 \\
 
93
  --overwrite False
94
 
95
  # Run the tasks in the specified task type, domains, and languages
96
+ python run_air_benchmark.py \\
97
  --task_types qa \\
98
  --domains wiki web \\
99
  --languages en \\
100
  --output_dir ./search_results \\
101
  --encoder BAAI/bge-m3 \\
 
102
  --reranker BAAI/bge-reranker-v2-m3 \\
 
103
  --search_top_k 1000 \\
104
  --rerank_top_k 100 \\
105
  --max_query_length 512 \\