feat: add model configs

.gitignore

@@ -0,0 +1 @@
+*.ckpt

models/orthrus_base_4_track/data_config.json

@@ -0,0 +1 @@
+{"transcript_save_dir": "/data1/morrisq/ian/rna_contrast/transcripts", "n_tracks": 6, "proportion_to_mask": 0.15, "short_transcript_map_path": "/data1/morrisq/ian/rna_contrast/maps/COMB-eutheria-splice_all_basic_shorter.pkl", "long_transcript_map_path": "/data1/morrisq/ian/rna_contrast/maps/COMB-eutheria-splice_all_basic_longer.pkl"}

models/orthrus_base_4_track/lp_results/homo_lp_results_ssm_6t_3_256_lr0.001_wd1e-05_mask0.15_splice_all_basic_eutheria_epoch=18-step=20000.ckpt_ridge.csv

@@ -0,0 +1,13 @@
+dataset,run_name,ckpt_name,lin_model,metric,split,mean_value,std_value
+go_mf,ssm_6t_3_256_lr0.001_wd1e-05_mask0.15_splice_all_basic_eutheria,epoch=18-step=20000.ckpt,logistic,val_roc_auc,val,0.8478578174637246,
+go_mf,ssm_6t_3_256_lr0.001_wd1e-05_mask0.15_splice_all_basic_eutheria,epoch=18-step=20000.ckpt,logistic,val_auprc,val,0.5095139671888503,
+mrfp,ssm_6t_3_256_lr0.001_wd1e-05_mask0.15_splice_all_basic_eutheria,epoch=18-step=20000.ckpt,ridge,val_loss,val,0.3546131256685185,
+mrfp,ssm_6t_3_256_lr0.001_wd1e-05_mask0.15_splice_all_basic_eutheria,epoch=18-step=20000.ckpt,ridge,val_r,val,0.5707018777098897,
+mrl,ssm_6t_3_256_lr0.001_wd1e-05_mask0.15_splice_all_basic_eutheria,epoch=18-step=20000.ckpt,ridge,val_loss,val,0.7273312686835766,
+mrl,ssm_6t_3_256_lr0.001_wd1e-05_mask0.15_splice_all_basic_eutheria,epoch=18-step=20000.ckpt,ridge,val_r,val,0.41711144859054816,
+protein_loc,ssm_6t_3_256_lr0.001_wd1e-05_mask0.15_splice_all_basic_eutheria,epoch=18-step=20000.ckpt,logistic,val_roc_auc,val,0.8337499171489103,
+protein_loc,ssm_6t_3_256_lr0.001_wd1e-05_mask0.15_splice_all_basic_eutheria,epoch=18-step=20000.ckpt,logistic,val_auprc,val,0.3868738259700707,
+hl_human,ssm_6t_3_256_lr0.001_wd1e-05_mask0.15_splice_all_basic_eutheria,epoch=18-step=20000.ckpt,ridge,val_loss,val,0.5510596183223497,
+hl_human,ssm_6t_3_256_lr0.001_wd1e-05_mask0.15_splice_all_basic_eutheria,epoch=18-step=20000.ckpt,ridge,val_r,val,0.6528813736053247,
+hl_mouse,ssm_6t_3_256_lr0.001_wd1e-05_mask0.15_splice_all_basic_eutheria,epoch=18-step=20000.ckpt,ridge,val_loss,val,0.5930341784262118,
+hl_mouse,ssm_6t_3_256_lr0.001_wd1e-05_mask0.15_splice_all_basic_eutheria,epoch=18-step=20000.ckpt,ridge,val_r,val,0.6290704100935987,

models/orthrus_base_4_track/model_config.json

@@ -0,0 +1 @@
+{"model_class": "ssm", "ssm_model_dim": 256, "ssm_n_layers": 3, "n_tracks": 6, "predict_splice_codon": false, "predict_masked": false}

models/orthrus_base_4_track/optimizer_config.json

@@ -0,0 +1 @@
+{"model_lr": 0.001, "model_weight_decay": 1e-05, "projection_head_lr": 0.001, "projection_head_weight_decay": 0.0001, "max_weight_decay": 0.0001, "clipnorm": 5, "warmup_steps": 2000.0}

models/orthrus_base_4_track/projector_config.json

@@ -0,0 +1 @@
+{"projection_body": 512, "projection_head_size": 128, "projection_norm_type": "batchnorm", "representation_dim": 256}

models/orthrus_base_4_track/train_config.json

@@ -0,0 +1 @@
+{"rand_seed": 0, "number_steps": 20000, "mixed_precision": true, "wandb_run_dir": "/data1/morrisq/ian/rna_contrast/grid_runs", "split_by_length": true, "gpu_batch_sizes": [40, 240, 40], "mask_prop": 0.0, "note": ""}

models/orthrus_large_6_track/data_config.json

@@ -0,0 +1 @@
+{"transcript_save_dir": "/data1/morrisq/ian/rna_contrast/transcripts", "n_tracks": 6, "proportion_to_mask": 0.15, "short_transcript_map_path": "/data1/morrisq/ian/rna_contrast/maps/COMB-eutheria-splice_all_basic_shorter.pkl", "long_transcript_map_path": "/data1/morrisq/ian/rna_contrast/maps/COMB-eutheria-splice_all_basic_longer.pkl"}

models/orthrus_large_6_track/lp_results/homo_lp_results_ssm_6_512_lr0.001_wd5e-05_mask0.15_seed0_splice_all_basic_eutheria_gene-dict_epoch=22-step=20000.ckpt.csv

@@ -0,0 +1,13 @@
+dataset,run_name,ckpt_name,lin_model,metric,split,mean_value,std_value
+go_mf,ssm_6_512_lr0.001_wd5e-05_mask0.15_seed0_splice_all_basic_eutheria_gene-dict,epoch=22-step=20000.ckpt,logistic,val_roc_auc,val,0.8689343127995597,
+go_mf,ssm_6_512_lr0.001_wd5e-05_mask0.15_seed0_splice_all_basic_eutheria_gene-dict,epoch=22-step=20000.ckpt,logistic,val_auprc,val,0.579792614711837,
+mrfp,ssm_6_512_lr0.001_wd5e-05_mask0.15_seed0_splice_all_basic_eutheria_gene-dict,epoch=22-step=20000.ckpt,linear,val_loss,val,0.4566991925239563,
+mrfp,ssm_6_512_lr0.001_wd5e-05_mask0.15_seed0_splice_all_basic_eutheria_gene-dict,epoch=22-step=20000.ckpt,linear,val_r,val,0.5779721134226916,
+mrl,ssm_6_512_lr0.001_wd5e-05_mask0.15_seed0_splice_all_basic_eutheria_gene-dict,epoch=22-step=20000.ckpt,linear,val_loss,val,0.7451710985674694,
+mrl,ssm_6_512_lr0.001_wd5e-05_mask0.15_seed0_splice_all_basic_eutheria_gene-dict,epoch=22-step=20000.ckpt,linear,val_r,val,0.4120694362754084,
+protein_loc,ssm_6_512_lr0.001_wd5e-05_mask0.15_seed0_splice_all_basic_eutheria_gene-dict,epoch=22-step=20000.ckpt,logistic,val_roc_auc,val,0.8369348139269305,
+protein_loc,ssm_6_512_lr0.001_wd5e-05_mask0.15_seed0_splice_all_basic_eutheria_gene-dict,epoch=22-step=20000.ckpt,logistic,val_auprc,val,0.3884790568417996,
+hl_human,ssm_6_512_lr0.001_wd5e-05_mask0.15_seed0_splice_all_basic_eutheria_gene-dict,epoch=22-step=20000.ckpt,linear,val_loss,val,0.5382314324378967,
+hl_human,ssm_6_512_lr0.001_wd5e-05_mask0.15_seed0_splice_all_basic_eutheria_gene-dict,epoch=22-step=20000.ckpt,linear,val_r,val,0.6649900713129258,
+hl_mouse,ssm_6_512_lr0.001_wd5e-05_mask0.15_seed0_splice_all_basic_eutheria_gene-dict,epoch=22-step=20000.ckpt,linear,val_loss,val,0.5822129845619202,
+hl_mouse,ssm_6_512_lr0.001_wd5e-05_mask0.15_seed0_splice_all_basic_eutheria_gene-dict,epoch=22-step=20000.ckpt,linear,val_r,val,0.6391530855147886,

models/orthrus_large_6_track/model_config.json

@@ -0,0 +1 @@
+{"model_class": "ssm", "ssm_model_dim": 512, "ssm_n_layers": 6}

models/orthrus_large_6_track/optimizer_config.json

@@ -0,0 +1 @@
+{"model_lr": 0.001, "model_weight_decay": 5e-05, "projection_head_lr": 0.001, "projection_head_weight_decay": 0.0001, "max_weight_decay": -1, "clipnorm": 5, "warmup_steps": 2000.0}

models/orthrus_large_6_track/projector_config.json

@@ -0,0 +1 @@
+{"projection_body": 512, "projection_head_size": 128, "projection_norm_type": "batchnorm", "representation_dim": 512}

models/orthrus_large_6_track/train_config.json

@@ -0,0 +1 @@
+{"rand_seed": 0, "number_steps": 20000, "mixed_precision": true, "wandb_run_dir": "/data1/morrisq/ian/rna_contrast/runs", "split_by_length": true, "gpu_batch_sizes": [50, 150, 50], "note": "gene-dict"}