(self)
| 38 | |
| 39 | class DataSelectionPipeline(): |
| 40 | def __init__(self): |
| 41 | self.storage = FileStorage( |
| 42 | first_entry_file_name="./dataflow/example/image_selection/prompts.jsonl", |
| 43 | cache_path="./cache_local", |
| 44 | file_name_prefix="dataflow_cache_step", |
| 45 | cache_type="jsonl" |
| 46 | ) |
| 47 | |
| 48 | self.rollout = VLMModelActor.remote( |
| 49 | "Qwen/Qwen2.5-VL-7B-Instruct" |
| 50 | ) |
| 51 | |
| 52 | self.verifier = LLMModelActor.remote( |
| 53 | "opencompass/CompassVerifier-3B" |
| 54 | ) |
| 55 | |
| 56 | self.rule_base_filter = RuleBaseFilter() |
| 57 | |
| 58 | self.clipscore_filter = CLIPScoreFilter( |
| 59 | model_name = 'openai/clip-vit-large-patch14-336', |
| 60 | keep_ratio=0.9, |
| 61 | ) |
| 62 | |
| 63 | self.deduplication_filter = ImageDeduplicateFilter( |
| 64 | model_name = 'openai/clip-vit-large-patch14-336', |
| 65 | threshold=0.95, |
| 66 | ) |
| 67 | |
| 68 | self.knn_similarity_filter = KNNSimilarityFilter( |
| 69 | model_name = 'openai/clip-vit-large-patch14-336', |
| 70 | k_neighbors=5, |
| 71 | keep_ratio=0.90, |
| 72 | ) |
| 73 | |
| 74 | # self.datatailor_filter = DataTailorFilter( |
| 75 | # model_name = 'Qwen/Qwen2.5-VL-7B-Instruct', |
| 76 | # keep_ratio=0.90 |
| 77 | # ) |
| 78 | |
| 79 | self.vision_dependent_filter = VisionDependentFilter( |
| 80 | rollout = self.rollout, |
| 81 | verifier = self.verifier |
| 82 | ) |
| 83 | |
| 84 | self.failrate_filter = FailRateFilter( |
| 85 | rollout = self.rollout, |
| 86 | verifier = self.verifier |
| 87 | ) |
| 88 | |
| 89 | |
| 90 | def forward(self): |
nothing calls this directly
no test coverage detected