| { | |
| "arch": { | |
| "vocab_size": 24217, | |
| "text_dim": 1536, | |
| "d_model": 384, | |
| "d_gene": 10, | |
| "d_hit": 64, | |
| "nhead": 2, | |
| "num_layers": 3, | |
| "dim_feedforward": 1024, | |
| "dropout": 0.0, | |
| "encoder_type": "transformer", | |
| "bert_model": "answerdotai/ModernBERT-base", | |
| "freeze_bert": false, | |
| "bert_pool": "cls", | |
| "text_max_tokens": 2048, | |
| "use_description": true, | |
| "freeze_factors": false, | |
| "disable_gene_bias": false, | |
| "variational": false | |
| }, | |
| "text_backend": "azure", | |
| "text_model": "text-embedding-3-small", | |
| "encoder_type": "transformer", | |
| "bert_model": "answerdotai/ModernBERT-base", | |
| "bert_pool": "cls", | |
| "text_max_tokens": 2048, | |
| "rl": { | |
| "init_checkpoint": "output/rankers/gf-bpmf-train-hits", | |
| "from_scratch": false, | |
| "aux_hit_coef": 0.0, | |
| "disable_gene_bias": false, | |
| "epochs": 100, | |
| "group_size": 8, | |
| "n_steps": 10, | |
| "batch_size": 100, | |
| "gamma": 1.0, | |
| "temperature": 1.0, | |
| "kl_coef": 0.0, | |
| "ent_coef": 0.01, | |
| "terminal_coef": 1.0, | |
| "reward_mode": "context_delta", | |
| "ctx_reset_epochs": 25, | |
| "adv_std_floor": 1e-06, | |
| "adv_clip": 0.0, | |
| "lr": 1e-05, | |
| "bert_lr": 1e-06, | |
| "rl_max_context": 1024, | |
| "seed": 19, | |
| "train_screen_set": "public_train", | |
| "eval_screen_set": "public_validation", | |
| "wandb_group": null, | |
| "wandb_tags": [ | |
| "rl", | |
| "gf", | |
| "full_genome" | |
| ], | |
| "warm_start": { | |
| "glm_train": null, | |
| "n": 0, | |
| "eval_dir": null, | |
| "eval_prefix": "sweep-ad6af08f-", | |
| "eval_warm_n": 0 | |
| } | |
| } | |
| } |