bge-m3-zbiory_urzedowe-partial

Bi-encoder model trained on paraphrases and new generated questions based on subset from datasets:

• ezd-ir
• ezd-qa
• opi-urzedowe

Context length was cut to 1024 tokens.

<!--- - original training name: bge-m3-ft-3 - wandb: https://wandb.ai/tdolega/pllum_retrievers_training/runs/r0qms5ns/overview - training params: { "lr": { "desc": null, "value": 0.00001 }, "_wandb": { "desc": null, "value": { "t": { "1": [ 1, 5, 11, 49, 53, 55, 75 ], "2": [ 1, 5, 11, 49, 53, 55, 75 ], "3": [ 2, 13, 16, 23 ], "4": "3.11.9", "5": "0.17.0", "6": "4.40.2", "8": [ 5 ], "13": "linux-x86_64" }, "framework": "huggingface", "start_time": 1716936164, "cli_version": "0.17.0", "is_jupyter_run": false, "python_version": "3.11.9", "is_kaggle_kernel": false, "huggingface_version": "4.40.2" } }, "loss_fn": { "desc": null, "value": "CachedMultipleNegativesRankingLoss" }, "use_amp": { "desc": null, "value": true }, "n_epochs": { "desc": null, "value": 4 }, "output_dir": { "desc": null, "value": "models" }, "test_steps": { "desc": null, "value": 5000 }, "input_model": { "desc": null, "value": "BAAI/bge-m3" }, "output_name": { "desc": null, "value": "bge-m3-ft-3" }, "lr_scheduler": { "desc": null, "value": "linear" }, "query_prefix": { "desc": null, "value": "" }, "weight_decay": { "desc": null, "value": 0.01 }, "passages_file": { "desc": null, "value": "data/zbiory_urzedowe/splitted/passages.jsonl.gz" }, "max_seq_length": { "desc": null, "value": 1024 }, "warmup_percent": { "desc": null, "value": 0.05 }, "mini_batch_size": { "desc": null, "value": 12 }, "test_batch_size": { "desc": null, "value": 12 }, "test_sample_size": { "desc": null, "value": 512 }, "train_batch_size": { "desc": null, "value": 256 }, "test_queries_file": { "desc": null, "value": "data/zbiory_urzedowe/splitted/queries_test.jsonl" }, "train_queries_file": { "desc": null, "value": "data/zbiory_urzedowe/splitted/queries_train.jsonl" } } -->