EtashGuha commited on
Commit
844c4a5
·
verified ·
1 Parent(s): 5a0873d

Upload configuration file

Browse files
Files changed (1) hide show
  1. config/swe_gym.yaml +39 -25
config/swe_gym.yaml CHANGED
@@ -9,9 +9,22 @@ operators:
9
  type: function
10
  function: data_strategies.commons.uniform_sample_fixed
11
  function_config:
12
- num_samples: 10
13
  input_ids:
14
  - load_hf_dataset
 
 
 
 
 
 
 
 
 
 
 
 
 
15
  - id: get_unique_code_sources
16
  config:
17
  type: function
@@ -26,27 +39,28 @@ operators:
26
  repo_column: repo
27
  input_ids:
28
  - get_unique_code_sources
29
- # - id: compute_relevant_context
30
- # config:
31
- # type: generic_resource_function
32
- # sharded: true
33
- # num_shards: 2
34
- # function: data_strategies.SWEGym.utils.compute_relevant_context
35
- # input_dataset_map:
36
- # dataset: sample_dataset
37
- # dataset_with_code_files: get_files_for_repo
38
- # input_ids:
39
- # - get_files_for_repo
40
- # - sample_dataset
41
- # - id: bm25k
42
- # config:
43
- # type: generic_resource_function
44
- # sharded: true
45
- # num_shards: 32
46
- # memory: 200
47
- # num_cpus: 32
48
- # function: data_strategies.SWEGym.utils.bm25k
49
- # function_config:
50
- # num_docs: 5
51
- # input_ids:
52
- # - compute_relevant_context
 
 
9
  type: function
10
  function: data_strategies.commons.uniform_sample_fixed
11
  function_config:
12
+ num_samples: 1000
13
  input_ids:
14
  - load_hf_dataset
15
+ - id: load_hf_dataset_2
16
+ config:
17
+ type: hf_source
18
+ dataset: SWE-Gym/SWE-Gym-Raw
19
+ split: train
20
+ - id: sample_dataset_2
21
+ config:
22
+ type: function
23
+ function: data_strategies.commons.uniform_sample_fixed
24
+ function_config:
25
+ num_samples: 1000
26
+ input_ids:
27
+ - load_hf_dataset_2
28
  - id: get_unique_code_sources
29
  config:
30
  type: function
 
39
  repo_column: repo
40
  input_ids:
41
  - get_unique_code_sources
42
+ - id: compute_relevant_context
43
+ config:
44
+ type: generic_resource_function
45
+ sharded: true
46
+ num_shards: 32
47
+ memory: 200
48
+ function: data_strategies.SWEGym.utils.compute_relevant_context
49
+ input_dataset_map:
50
+ dataset: sample_dataset_2
51
+ dataset_with_code_files: get_files_for_repo
52
+ input_ids:
53
+ - sample_dataset_2
54
+ - get_files_for_repo
55
+ - id: bm25k
56
+ config:
57
+ type: generic_resource_function
58
+ sharded: true
59
+ num_shards: 32
60
+ memory: 200
61
+ num_cpus: 32
62
+ function: data_strategies.SWEGym.utils.bm25k
63
+ function_config:
64
+ num_docs: 5
65
+ input_ids:
66
+ - compute_relevant_context