codemix-retriever / codemix_training.json
roshan9136's picture
CodeMixESC retriever
97cb72a verified
Raw History Blame Contribute Delete
1.82 kB
{
"config": {
"base_model": "sentence-transformers/paraphrase-multilingual-mpnet-base-v2",
"pairs_dir": "data/pairs",
"out_dir": "models/codemix-retriever",
"ckpt_dir": "models/ckpt-e1",
"log": "results/retrieval/train_log.json",
"epochs": 1.0,
"batch_size": 32,
"lr": 2e-05,
"warmup_ratio": 0.1,
"weight_decay": 0.01,
"scale": 20.0,
"max_seq_length": 128,
"freeze_embeddings": "auto",
"gradient_checkpointing": false,
"cached": false,
"mini_batch_size": 16,
"no_fp16": false,
"optim": "adamw_torch",
"evals_per_epoch": 4,
"save_total_limit": 2,
"eval_batch_size": 64,
"seed": 42,
"max_steps": 0,
"limit": 0,
"bank_limit": 0,
"resume": true,
"keep_checkpoints": false,
"tau": 0.05,
"k": 10,
"eval_steps": 132,
"steps_per_epoch": 530,
"total_steps": 530,
"fp16": true,
"batch_sampler": "no_duplicates (shuffled, see docstring)",
"loss": "MultipleNegativesRankingLoss"
},
"data": {
"train_pairs": 16951,
"val_pairs": 702,
"dev_queries": 302,
"bank": 13334,
"train_by_source": {
"phinc": 12156,
"esconv": 4795
},
"dev_levels": [
"heavy",
"light"
],
"selection_metric": "dev"
},
"params": {
"total": 278043648,
"frozen": 192001536,
"trainable": 86042112,
"word_embeddings_frozen": true
},
"best_step": 264,
"final_eval": {
"dev_heavy_overlap@10": 0.5397350993377483,
"dev_heavy_p@10": 0.2867549668874172,
"dev_light_overlap@10": 0.7291390728476821,
"dev_light_p@10": 0.26953642384105964,
"dev_overlap@10": 0.6344370860927152,
"dev_p@10": 0.2781456953642384,
"dev_unique@10": 0.6374172185430463,
"dev_primary": 0.4562913907284768,
"val_acc@1": 0.8831908831908832,
"val_mrr@10": 0.9041384027495138,
"primary": 0.4562913907284768
},
"git_commit": "78e697fc1033ade75490f5e36b7f24840b4fa472"
}