Release the 25 Varformer paper models

README.md

@@ -0,0 +1,57 @@
+---
+license: cc-by-nc-sa-4.0
+library_name: pytorch
+tags:
+  - biology
+  - genomics
+  - drug-discovery
+  - target-prioritization
+  - variant-prioritization
+---
+
+# Varformer
+
+Varformer is a multimodal transformer for population-scale drug target and variant prioritization. It learns contextual representations of the missense variants observed in a population and integrates them with gene-level biological features through cross-modal attention. It is trained with a non-negative positive-unlabeled objective for the gene score, and an allele-frequency objective for a per-variant score.
+
+Code, documentation and installation: https://github.com/aaronwtr/Varformer
+
+## Models in this repository
+
+The 25 models of the paper. Each was trained with five seeds (13, 23, 42, 85, 482), with the variant head (`variant_loss_weight` 0.1, pooled binary cross-entropy) and 32-bit precision.
+
+| Folder | Cohort | Variant source | Gene inputs | Use in the paper |
+|---|---|---|---|---|
+| `afr/` | African / African American | gnomAD v4.1 | without `geneticConstraint` | gene prioritization |
+| `amr/` | Admixed American | gnomAD v4.1 | without `geneticConstraint` | gene prioritization |
+| `nfe/` | Non-Finnish European | gnomAD v4.1 | without `geneticConstraint` | gene prioritization |
+| `sas/` | South Asian | Genes & Health | without `geneticConstraint` | gene prioritization |
+| `sas-gnh/` | South Asian | Genes & Health | with `geneticConstraint` | variant scores in the Genes & Health analyses |
+
+Each model has two files:
+
+- `seed<N>.safetensors`: the model weights only. The files contain no training data, no genotypes and no individual-level information.
+- `seed<N>.json`: the training settings, the input dimensions, the class prior, the decision threshold, the selected epoch, the validation Spearman correlation, and SHA-256 checksums of the weights and of the source checkpoint.
+
+`index.json` lists all 25 models.
+
+## Use
+
+```python
+from varformer import Varformer
+
+model = Varformer.from_pretrained("nfe", seed="best")
+predictions = model.predict(genes=["ENSG00000141510"])
+variant_scores = model.score_variants(genes=["ENSG00000141510"])
+```
+
+`from_pretrained` checks the weights against their checksum. The model inputs are built from local copies of the input data (gnomAD or Genes & Health variants, AlphaMissense scores, Open Targets, Human Protein Atlas and GTEx features); see the data sources in the code repository. Genes & Health data are available only to approved researchers.
+
+The gene score ranks genes by target-prioritization evidence; it is not a calibrated probability of clinical success. The variant score ranks variants for research prioritization; it is not validated for clinical variant classification.
+
+## License
+
+The weights are released under [CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/): attribution, non-commercial use only, and the same license for derived works. The models take AlphaMissense scores as input, and AlphaMissense is released under the same license. The code is MIT-licensed. For commercial use, train a model with another pathogenicity predictor; the code repository describes how.
+
+## Citation
+
+The paper citation will be added when the preprint is available. Until then, please cite the software release described in the code repository.

afr/seed13.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.08368786966898413,
+  "decision_threshold": 0.7535547614097595,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 527,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "afr",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 13,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.7535547373130804,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8379686,
+  "population": "afr",
+  "seed": 13,
+  "selected_epoch": 23,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "cd70d456eee7b6da5aafd7243daeac5c72ad3afdac2025f458bda4e002280ccd",
+  "validation_spearman": 0.41,
+  "variant_head": true,
+  "weights_sha256": "4c79b8b76c3d5b8cade53308e3547e0969e0f62d1c85b3d044fdce88b354385a"
+}

afr/seed13.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4c79b8b76c3d5b8cade53308e3547e0969e0f62d1c85b3d044fdce88b354385a
+size 33527716

afr/seed23.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.08368786966898413,
+  "decision_threshold": 0.5682888627052307,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 527,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "afr",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 23,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.5682888836537554,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8379686,
+  "population": "afr",
+  "seed": 23,
+  "selected_epoch": 16,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "3ba8ce07d1ddc7b2c94dc4624294cae47c0fb636f05fc91d66887e224d084df5",
+  "validation_spearman": 0.38,
+  "variant_head": true,
+  "weights_sha256": "a0822954a24ce0da917436847a7531fbac10cf3a5b4caf47112b96e194ec0346"
+}

afr/seed23.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0822954a24ce0da917436847a7531fbac10cf3a5b4caf47112b96e194ec0346
+size 33527716

afr/seed42.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.08368786966898413,
+  "decision_threshold": 0.5650092959403992,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 527,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "afr",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 42,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.5650092758701416,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8379686,
+  "population": "afr",
+  "seed": 42,
+  "selected_epoch": 28,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "6b420977f96351c6fe627ed22c738276960f20025ce4f80fe4afaf40e66cf41f",
+  "validation_spearman": 0.39,
+  "variant_head": true,
+  "weights_sha256": "6be35ab0c05403c85408a61cf381d08a302fafb0288ee4c11cd8b3a8fa1d643b"
+}

afr/seed42.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6be35ab0c05403c85408a61cf381d08a302fafb0288ee4c11cd8b3a8fa1d643b
+size 33527716

afr/seed482.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.08368786966898413,
+  "decision_threshold": 0.531441330909729,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 527,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "afr",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 482,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.5314413357773077,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8379686,
+  "population": "afr",
+  "seed": 482,
+  "selected_epoch": 30,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "a5e14abe23f9e43c9edb131eaaaf794d26f7143d13570a35df9de67a312401e3",
+  "validation_spearman": 0.41,
+  "variant_head": true,
+  "weights_sha256": "efe418a39c330fd0887f1024692a793dac9b150e5535b484a895e21bc77073f3"
+}

afr/seed482.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:efe418a39c330fd0887f1024692a793dac9b150e5535b484a895e21bc77073f3
+size 33527716

afr/seed85.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.08368786966898413,
+  "decision_threshold": 0.45899805426597595,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 527,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "afr",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 85,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.45899805593901283,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8379686,
+  "population": "afr",
+  "seed": 85,
+  "selected_epoch": 20,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "75eef825c2597af5aa88fa317cf8957c7e9aecb391aab086a9bf2f24a2f0d46a",
+  "validation_spearman": 0.38,
+  "variant_head": true,
+  "weights_sha256": "2936c16df39cf65822ea29deaea43078f9e4bec601ca9b0662ba2a6c1640fd9d"
+}

afr/seed85.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2936c16df39cf65822ea29deaea43078f9e4bec601ca9b0662ba2a6c1640fd9d
+size 33527716

amr/seed13.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.08360128617363344,
+  "decision_threshold": 0.5641329884529114,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 527,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "amr",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 13,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.5641329796367903,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8379686,
+  "population": "amr",
+  "seed": 13,
+  "selected_epoch": 26,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "956a348cb79f0a3534a0a7f526d4e0367f988315017649e3e99a4a8ab75d027a",
+  "validation_spearman": 0.4,
+  "variant_head": true,
+  "weights_sha256": "dcecf4c1f6bbd20e984e9484a8c58daef36814517f6fd59a2597504e99e7b239"
+}

amr/seed13.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dcecf4c1f6bbd20e984e9484a8c58daef36814517f6fd59a2597504e99e7b239
+size 33527716

amr/seed23.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.08360128617363344,
+  "decision_threshold": 0.5932214856147766,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 527,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "amr",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 23,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.5932215043969471,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8379686,
+  "population": "amr",
+  "seed": 23,
+  "selected_epoch": 29,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "90b9174033575013f3376e752b267513006e76cb26ba652a27a9321618a0f924",
+  "validation_spearman": 0.39,
+  "variant_head": true,
+  "weights_sha256": "6445517f741d337a2b6fb1818a945db28973268a41d77cf4fa445a1806b51caf"
+}

amr/seed23.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6445517f741d337a2b6fb1818a945db28973268a41d77cf4fa445a1806b51caf
+size 33527716

amr/seed42.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.08360128617363344,
+  "decision_threshold": 0.6023826003074646,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 527,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "amr",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 42,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.6023825853583891,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8379686,
+  "population": "amr",
+  "seed": 42,
+  "selected_epoch": 40,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "eb871dbf93b71b98fb357facd44ab82e761266a8a0900123759098a2dcf2dcbc",
+  "validation_spearman": 0.39,
+  "variant_head": true,
+  "weights_sha256": "f0d290cb431bbe377bbd5144f382a6ff5bab47b559c730bc8a97e25aa678ae6a"
+}

amr/seed42.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0d290cb431bbe377bbd5144f382a6ff5bab47b559c730bc8a97e25aa678ae6a
+size 33527716

amr/seed482.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.08360128617363344,
+  "decision_threshold": 0.8855880498886108,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 527,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "amr",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 482,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.8855880675208528,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8379686,
+  "population": "amr",
+  "seed": 482,
+  "selected_epoch": 37,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "b4a6e256e98a67436fd5b8a91d617822dfb44e446976750c1987d94155a235eb",
+  "validation_spearman": 0.41,
+  "variant_head": true,
+  "weights_sha256": "f5ab4e1eb2ffe5125628709db8a43317ec665abbbbcb3d68a2be99c465f89f68"
+}

amr/seed482.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5ab4e1eb2ffe5125628709db8a43317ec665abbbbcb3d68a2be99c465f89f68
+size 33527716

amr/seed85.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.08360128617363344,
+  "decision_threshold": 0.6900997757911682,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 527,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "amr",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 85,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.6900997972565062,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8379686,
+  "population": "amr",
+  "seed": 85,
+  "selected_epoch": 30,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "31237492810c1b61e9fce7b88abd456d1170733a8e86838335c749c019f786a7",
+  "validation_spearman": 0.4,
+  "variant_head": true,
+  "weights_sha256": "24220f19f9c837613afd9c4b9bc493ed51715950f22f308bb09aebb4cc8c32b0"
+}

amr/seed85.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24220f19f9c837613afd9c4b9bc493ed51715950f22f308bb09aebb4cc8c32b0
+size 33527716

index.json

@@ -0,0 +1,317 @@
+[
+  {
+    "file": "afr/seed13.safetensors",
+    "population": "afr",
+    "seed": 13,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.41,
+    "selected_epoch": 23,
+    "weights_sha256": "4c79b8b76c3d5b8cade53308e3547e0969e0f62d1c85b3d044fdce88b354385a",
+    "source_checkpoint_sha256": "cd70d456eee7b6da5aafd7243daeac5c72ad3afdac2025f458bda4e002280ccd",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "afr/seed23.safetensors",
+    "population": "afr",
+    "seed": 23,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.38,
+    "selected_epoch": 16,
+    "weights_sha256": "a0822954a24ce0da917436847a7531fbac10cf3a5b4caf47112b96e194ec0346",
+    "source_checkpoint_sha256": "3ba8ce07d1ddc7b2c94dc4624294cae47c0fb636f05fc91d66887e224d084df5",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "afr/seed42.safetensors",
+    "population": "afr",
+    "seed": 42,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.39,
+    "selected_epoch": 28,
+    "weights_sha256": "6be35ab0c05403c85408a61cf381d08a302fafb0288ee4c11cd8b3a8fa1d643b",
+    "source_checkpoint_sha256": "6b420977f96351c6fe627ed22c738276960f20025ce4f80fe4afaf40e66cf41f",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "afr/seed85.safetensors",
+    "population": "afr",
+    "seed": 85,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.38,
+    "selected_epoch": 20,
+    "weights_sha256": "2936c16df39cf65822ea29deaea43078f9e4bec601ca9b0662ba2a6c1640fd9d",
+    "source_checkpoint_sha256": "75eef825c2597af5aa88fa317cf8957c7e9aecb391aab086a9bf2f24a2f0d46a",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "afr/seed482.safetensors",
+    "population": "afr",
+    "seed": 482,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.41,
+    "selected_epoch": 30,
+    "weights_sha256": "efe418a39c330fd0887f1024692a793dac9b150e5535b484a895e21bc77073f3",
+    "source_checkpoint_sha256": "a5e14abe23f9e43c9edb131eaaaf794d26f7143d13570a35df9de67a312401e3",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "amr/seed13.safetensors",
+    "population": "amr",
+    "seed": 13,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.4,
+    "selected_epoch": 26,
+    "weights_sha256": "dcecf4c1f6bbd20e984e9484a8c58daef36814517f6fd59a2597504e99e7b239",
+    "source_checkpoint_sha256": "956a348cb79f0a3534a0a7f526d4e0367f988315017649e3e99a4a8ab75d027a",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "amr/seed23.safetensors",
+    "population": "amr",
+    "seed": 23,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.39,
+    "selected_epoch": 29,
+    "weights_sha256": "6445517f741d337a2b6fb1818a945db28973268a41d77cf4fa445a1806b51caf",
+    "source_checkpoint_sha256": "90b9174033575013f3376e752b267513006e76cb26ba652a27a9321618a0f924",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "amr/seed42.safetensors",
+    "population": "amr",
+    "seed": 42,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.39,
+    "selected_epoch": 40,
+    "weights_sha256": "f0d290cb431bbe377bbd5144f382a6ff5bab47b559c730bc8a97e25aa678ae6a",
+    "source_checkpoint_sha256": "eb871dbf93b71b98fb357facd44ab82e761266a8a0900123759098a2dcf2dcbc",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "amr/seed85.safetensors",
+    "population": "amr",
+    "seed": 85,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.4,
+    "selected_epoch": 30,
+    "weights_sha256": "24220f19f9c837613afd9c4b9bc493ed51715950f22f308bb09aebb4cc8c32b0",
+    "source_checkpoint_sha256": "31237492810c1b61e9fce7b88abd456d1170733a8e86838335c749c019f786a7",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "amr/seed482.safetensors",
+    "population": "amr",
+    "seed": 482,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.41,
+    "selected_epoch": 37,
+    "weights_sha256": "f5ab4e1eb2ffe5125628709db8a43317ec665abbbbcb3d68a2be99c465f89f68",
+    "source_checkpoint_sha256": "b4a6e256e98a67436fd5b8a91d617822dfb44e446976750c1987d94155a235eb",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "nfe/seed13.safetensors",
+    "population": "nfe",
+    "seed": 13,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.4,
+    "selected_epoch": 34,
+    "weights_sha256": "0c5de915aae59ad0b4ade78eeeff0039fb7666ced4840f5347005c474e29dd30",
+    "source_checkpoint_sha256": "a10dda732716d55944ce5402e6065cd3217d361448d8e4cb849ae6910c905d68",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "nfe/seed23.safetensors",
+    "population": "nfe",
+    "seed": 23,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.39,
+    "selected_epoch": 19,
+    "weights_sha256": "dd3fb603b602c686b3ad68908695c12a5bbe3d03efb54dc7301a036094434588",
+    "source_checkpoint_sha256": "1bf2dbc21482cace72b2f636d890f976f12a6e8f4ebc8a83d275cf8ab77233bd",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "nfe/seed42.safetensors",
+    "population": "nfe",
+    "seed": 42,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.4,
+    "selected_epoch": 40,
+    "weights_sha256": "98b7f98162a238e3525c2af28c4cb8aa54a4d6ce7f9e387d67235a5f89366a44",
+    "source_checkpoint_sha256": "734bc122f8f980211f7788d94c261828cae06c6c4ec869655b85bade0ef6a98f",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "nfe/seed85.safetensors",
+    "population": "nfe",
+    "seed": 85,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.39,
+    "selected_epoch": 29,
+    "weights_sha256": "acf777405901658bc1b7e9d071473d87ae49a21720bd3dcd20bb6f945049f371",
+    "source_checkpoint_sha256": "f588461020fc7b6a39ffe34575f493e3223f12b15d6caffc9498bc1aacc7175f",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "nfe/seed482.safetensors",
+    "population": "nfe",
+    "seed": 482,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.38,
+    "selected_epoch": 15,
+    "weights_sha256": "dd2b78a9c8bf08f08127ec5e8ca606ef5a460196686e17cd16ae4cda01b5283d",
+    "source_checkpoint_sha256": "fb5358431175a362baeee1ac49ef1e69c51c06cef9c4ddcedd945c101547f2f4",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "sas/seed13.safetensors",
+    "population": "sas",
+    "seed": 13,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.44,
+    "selected_epoch": 48,
+    "weights_sha256": "f3ed37f7c609b6ead69b442b0e0761a18c74336abf56788ebc0058953c158099",
+    "source_checkpoint_sha256": "e5cecb69c2ccb5bcd80c56aae1b4a84e8a379a69fb41f96a8d6b892ee46a9ce2",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "sas/seed23.safetensors",
+    "population": "sas",
+    "seed": 23,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.41,
+    "selected_epoch": 16,
+    "weights_sha256": "3b297d849b09f3e6d025351a62db91bae7d150ddcc6ffcf3bd0c2699b9fe1554",
+    "source_checkpoint_sha256": "88a42931d355344ed42d80cde5b02ff0ceb3787a5de56bfd212d522acd785602",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "sas/seed42.safetensors",
+    "population": "sas",
+    "seed": 42,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.41,
+    "selected_epoch": 38,
+    "weights_sha256": "ad938cb10f958bece96a881a1db4c00b6a07a16c5e1097d211d539ee6cda99f6",
+    "source_checkpoint_sha256": "f3e4d7f90f82ce87f7cb89d323fe41e4f7bb938879ec276aaf7d3e5730d2c880",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "sas/seed85.safetensors",
+    "population": "sas",
+    "seed": 85,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.42,
+    "selected_epoch": 28,
+    "weights_sha256": "59e6f1f76b2c95457863a71d5e2e93f5b2a8dcdc23e152b207df0630da13c2ec",
+    "source_checkpoint_sha256": "f006488787655ba28636e87050c74acccaef9a12e8da296ccb8d23c84a325593",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "sas/seed482.safetensors",
+    "population": "sas",
+    "seed": 482,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "validation_spearman": 0.41,
+    "selected_epoch": 21,
+    "weights_sha256": "ebf099391138a2d98189cd6fcb7d6117d9d274674f8eeeef18aa5f3907d83f57",
+    "source_checkpoint_sha256": "5611864f162ff430d278fba27b80e852995b28be4e91843e8c503276f75d7c21",
+    "source_check": "evaluation_manifest"
+  },
+  {
+    "file": "sas-gnh/seed13.safetensors",
+    "population": "sas",
+    "seed": 13,
+    "gc_exclude_features": [],
+    "validation_spearman": 0.43,
+    "selected_epoch": 36,
+    "weights_sha256": "eda3c7c410005d6eea833aa36755176878ec4615390a7546f844d687b1c86062",
+    "source_checkpoint_sha256": "ff2adb781074ed57e02943fdc8ca774376271a638dccedffd99b61a4519e55eb",
+    "source_check": "run_manifest"
+  },
+  {
+    "file": "sas-gnh/seed23.safetensors",
+    "population": "sas",
+    "seed": 23,
+    "gc_exclude_features": [],
+    "validation_spearman": 0.41,
+    "selected_epoch": 37,
+    "weights_sha256": "9ff25a0fe7c7ecd54e00c1c66a654ff5161ef0330ebe6144d64a9374c0df3b66",
+    "source_checkpoint_sha256": "729944b5fde327e3ab64935d5e129385fb96ae0d711a428cd847941b5ff848ca",
+    "source_check": "run_manifest"
+  },
+  {
+    "file": "sas-gnh/seed42.safetensors",
+    "population": "sas",
+    "seed": 42,
+    "gc_exclude_features": [],
+    "validation_spearman": 0.4,
+    "selected_epoch": 29,
+    "weights_sha256": "92022a8a561fb8dba521ffd88ab010df41fce23ec848472bb68720e3db311b77",
+    "source_checkpoint_sha256": "e535bef88980c2303ffa2e95925bbb4d399585020e2c190f7281061665fdb89d",
+    "source_check": "run_manifest"
+  },
+  {
+    "file": "sas-gnh/seed85.safetensors",
+    "population": "sas",
+    "seed": 85,
+    "gc_exclude_features": [],
+    "validation_spearman": 0.41,
+    "selected_epoch": 28,
+    "weights_sha256": "55c37906838a195769e3a010a6adf3732deb9dc3a91ca0de4d7f0d428a62bc6a",
+    "source_checkpoint_sha256": "31fe3d9549eab5ae3041ca244fc3865b7b3351f1f7b196bf59811e76f020b011",
+    "source_check": "run_manifest"
+  },
+  {
+    "file": "sas-gnh/seed482.safetensors",
+    "population": "sas",
+    "seed": 482,
+    "gc_exclude_features": [],
+    "validation_spearman": 0.41,
+    "selected_epoch": 30,
+    "weights_sha256": "75c1187f3e93af73579cd327f044276af41adeca9b209de005b0c2b267a21c7b",
+    "source_checkpoint_sha256": "2ae1e885532c107625270a2b9bcf85d4ca4d22dc17e37b02c9a4021a604c5d65",
+    "source_check": "run_manifest"
+  }
+]

nfe/seed13.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.0814781160078752,
+  "decision_threshold": 0.6301044821739197,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 527,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "nfe",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 13,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.6301045029719057,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8379686,
+  "population": "nfe",
+  "seed": 13,
+  "selected_epoch": 34,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "a10dda732716d55944ce5402e6065cd3217d361448d8e4cb849ae6910c905d68",
+  "validation_spearman": 0.4,
+  "variant_head": true,
+  "weights_sha256": "0c5de915aae59ad0b4ade78eeeff0039fb7666ced4840f5347005c474e29dd30"
+}

nfe/seed13.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c5de915aae59ad0b4ade78eeeff0039fb7666ced4840f5347005c474e29dd30
+size 33527716

nfe/seed23.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.0814781160078752,
+  "decision_threshold": 0.6867796182632446,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 527,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "nfe",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 23,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.6867796320744065,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8379686,
+  "population": "nfe",
+  "seed": 23,
+  "selected_epoch": 19,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "1bf2dbc21482cace72b2f636d890f976f12a6e8f4ebc8a83d275cf8ab77233bd",
+  "validation_spearman": 0.39,
+  "variant_head": true,
+  "weights_sha256": "dd3fb603b602c686b3ad68908695c12a5bbe3d03efb54dc7301a036094434588"
+}

nfe/seed23.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd3fb603b602c686b3ad68908695c12a5bbe3d03efb54dc7301a036094434588
+size 33527716

nfe/seed42.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.0814781160078752,
+  "decision_threshold": 0.8719609975814819,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 527,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "nfe",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 42,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.8719609837432385,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8379686,
+  "population": "nfe",
+  "seed": 42,
+  "selected_epoch": 40,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "734bc122f8f980211f7788d94c261828cae06c6c4ec869655b85bade0ef6a98f",
+  "validation_spearman": 0.4,
+  "variant_head": true,
+  "weights_sha256": "98b7f98162a238e3525c2af28c4cb8aa54a4d6ce7f9e387d67235a5f89366a44"
+}

nfe/seed42.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98b7f98162a238e3525c2af28c4cb8aa54a4d6ce7f9e387d67235a5f89366a44
+size 33527716

nfe/seed482.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.0814781160078752,
+  "decision_threshold": 0.3766169250011444,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 527,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "nfe",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 482,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.37661692854871753,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8379686,
+  "population": "nfe",
+  "seed": 482,
+  "selected_epoch": 15,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "fb5358431175a362baeee1ac49ef1e69c51c06cef9c4ddcedd945c101547f2f4",
+  "validation_spearman": 0.38,
+  "variant_head": true,
+  "weights_sha256": "dd2b78a9c8bf08f08127ec5e8ca606ef5a460196686e17cd16ae4cda01b5283d"
+}

nfe/seed482.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd2b78a9c8bf08f08127ec5e8ca606ef5a460196686e17cd16ae4cda01b5283d
+size 33527716

nfe/seed85.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.0814781160078752,
+  "decision_threshold": 0.8027001619338989,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 527,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "nfe",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 85,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.8027001751492858,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8379686,
+  "population": "nfe",
+  "seed": 85,
+  "selected_epoch": 29,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "f588461020fc7b6a39ffe34575f493e3223f12b15d6caffc9498bc1aacc7175f",
+  "validation_spearman": 0.39,
+  "variant_head": true,
+  "weights_sha256": "acf777405901658bc1b7e9d071473d87ae49a21720bd3dcd20bb6f945049f371"
+}

nfe/seed85.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:acf777405901658bc1b7e9d071473d87ae49a21720bd3dcd20bb6f945049f371
+size 33527716

sas-gnh/seed13.json

@@ -0,0 +1,72 @@
+{
+  "class_prior": 0.08849664510669165,
+  "decision_threshold": 0.4418480694293976,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 10,
+    "num_features_go": 524,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "sas",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 13,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.4418480829839327,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8378182,
+  "population": "sas",
+  "seed": 13,
+  "selected_epoch": 36,
+  "source_check": "run_manifest",
+  "source_checkpoint_sha256": "ff2adb781074ed57e02943fdc8ca774376271a638dccedffd99b61a4519e55eb",
+  "validation_spearman": 0.43,
+  "variant_head": true,
+  "weights_sha256": "eda3c7c410005d6eea833aa36755176878ec4615390a7546f844d687b1c86062"
+}

sas-gnh/seed13.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eda3c7c410005d6eea833aa36755176878ec4615390a7546f844d687b1c86062
+size 33521700

sas-gnh/seed23.json

@@ -0,0 +1,72 @@
+{
+  "class_prior": 0.08849664510669165,
+  "decision_threshold": 0.7796894907951355,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 10,
+    "num_features_go": 524,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "sas",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 23,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.7796894906005755,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8378182,
+  "population": "sas",
+  "seed": 23,
+  "selected_epoch": 37,
+  "source_check": "run_manifest",
+  "source_checkpoint_sha256": "729944b5fde327e3ab64935d5e129385fb96ae0d711a428cd847941b5ff848ca",
+  "validation_spearman": 0.41,
+  "variant_head": true,
+  "weights_sha256": "9ff25a0fe7c7ecd54e00c1c66a654ff5161ef0330ebe6144d64a9374c0df3b66"
+}

sas-gnh/seed23.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ff25a0fe7c7ecd54e00c1c66a654ff5161ef0330ebe6144d64a9374c0df3b66
+size 33521700

sas-gnh/seed42.json

@@ -0,0 +1,72 @@
+{
+  "class_prior": 0.08849664510669165,
+  "decision_threshold": 0.6333062052726746,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 10,
+    "num_features_go": 524,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "sas",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 42,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.6333061968236093,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8378182,
+  "population": "sas",
+  "seed": 42,
+  "selected_epoch": 29,
+  "source_check": "run_manifest",
+  "source_checkpoint_sha256": "e535bef88980c2303ffa2e95925bbb4d399585020e2c190f7281061665fdb89d",
+  "validation_spearman": 0.4,
+  "variant_head": true,
+  "weights_sha256": "92022a8a561fb8dba521ffd88ab010df41fce23ec848472bb68720e3db311b77"
+}

sas-gnh/seed42.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:92022a8a561fb8dba521ffd88ab010df41fce23ec848472bb68720e3db311b77
+size 33521700

sas-gnh/seed482.json

@@ -0,0 +1,72 @@
+{
+  "class_prior": 0.08849664510669165,
+  "decision_threshold": 0.7612968683242798,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 10,
+    "num_features_go": 524,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "sas",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 482,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.7612968543445895,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8378182,
+  "population": "sas",
+  "seed": 482,
+  "selected_epoch": 30,
+  "source_check": "run_manifest",
+  "source_checkpoint_sha256": "2ae1e885532c107625270a2b9bcf85d4ca4d22dc17e37b02c9a4021a604c5d65",
+  "validation_spearman": 0.41,
+  "variant_head": true,
+  "weights_sha256": "75c1187f3e93af73579cd327f044276af41adeca9b209de005b0c2b267a21c7b"
+}

sas-gnh/seed482.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:75c1187f3e93af73579cd327f044276af41adeca9b209de005b0c2b267a21c7b
+size 33521700

sas-gnh/seed85.json

@@ -0,0 +1,72 @@
+{
+  "class_prior": 0.08849664510669165,
+  "decision_threshold": 0.2809520363807678,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 10,
+    "num_features_go": 524,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "sas",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 85,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.2809520276975065,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8378182,
+  "population": "sas",
+  "seed": 85,
+  "selected_epoch": 28,
+  "source_check": "run_manifest",
+  "source_checkpoint_sha256": "31fe3d9549eab5ae3041ca244fc3865b7b3351f1f7b196bf59811e76f020b011",
+  "validation_spearman": 0.41,
+  "variant_head": true,
+  "weights_sha256": "55c37906838a195769e3a010a6adf3732deb9dc3a91ca0de4d7f0d428a62bc6a"
+}

sas-gnh/seed85.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55c37906838a195769e3a010a6adf3732deb9dc3a91ca0de4d7f0d428a62bc6a
+size 33521700

sas/seed13.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.08849664510669165,
+  "decision_threshold": 0.6496922969818115,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 524,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "sas",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 13,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.6496922880823696,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8378150,
+  "population": "sas",
+  "seed": 13,
+  "selected_epoch": 48,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "e5cecb69c2ccb5bcd80c56aae1b4a84e8a379a69fb41f96a8d6b892ee46a9ce2",
+  "validation_spearman": 0.44,
+  "variant_head": true,
+  "weights_sha256": "f3ed37f7c609b6ead69b442b0e0761a18c74336abf56788ebc0058953c158099"
+}

sas/seed13.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f3ed37f7c609b6ead69b442b0e0761a18c74336abf56788ebc0058953c158099
+size 33521572

sas/seed23.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.08849664510669165,
+  "decision_threshold": 0.8452841639518738,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 524,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "sas",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 23,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.8452841911906526,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8378150,
+  "population": "sas",
+  "seed": 23,
+  "selected_epoch": 16,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "88a42931d355344ed42d80cde5b02ff0ceb3787a5de56bfd212d522acd785602",
+  "validation_spearman": 0.41,
+  "variant_head": true,
+  "weights_sha256": "3b297d849b09f3e6d025351a62db91bae7d150ddcc6ffcf3bd0c2699b9fe1554"
+}

sas/seed23.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3b297d849b09f3e6d025351a62db91bae7d150ddcc6ffcf3bd0c2699b9fe1554
+size 33521572

sas/seed42.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.08849664510669165,
+  "decision_threshold": 0.9446958899497986,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 524,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "sas",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 42,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.9446958622065983,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8378150,
+  "population": "sas",
+  "seed": 42,
+  "selected_epoch": 38,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "f3e4d7f90f82ce87f7cb89d323fe41e4f7bb938879ec276aaf7d3e5730d2c880",
+  "validation_spearman": 0.41,
+  "variant_head": true,
+  "weights_sha256": "ad938cb10f958bece96a881a1db4c00b6a07a16c5e1097d211d539ee6cda99f6"
+}

sas/seed42.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad938cb10f958bece96a881a1db4c00b6a07a16c5e1097d211d539ee6cda99f6
+size 33521572

sas/seed482.json

@@ -0,0 +1,76 @@
+{
+  "class_prior": 0.08849664510669165,
+  "decision_threshold": 0.630003809928894,
+  "dimensions": {
+    "max_seq_len": 1024,
+    "num_features_gc": 9,
+    "num_features_go": 524,
+    "num_genes": 1024,
+    "num_mutations": 381
+  },
+  "gc_exclude_features": [
+    "geneticConstraint"
+  ],
+  "hyperparameters": {
+    "T0": 200,
+    "af_log10_floor": -6.0,
+    "batch_size": 128,
+    "d_model": 256,
+    "depth_cls_head": 4,
+    "dim_feedforward": 4096,
+    "dropout": 0.3,
+    "epochs": 100,
+    "gc_exclude_features": [
+      "geneticConstraint"
+    ],
+    "gc_width": 32,
+    "go_width": 512,
+    "grad_accum": null,
+    "gradient_clip_val": 1.0,
+    "gv_attn_dim": 256,
+    "include_open_targets_labels": true,
+    "lr_end": 1e-05,
+    "lr_start": 0.0001,
+    "mask_cross_attention_padding": true,
+    "max_seq_len": 1024,
+    "min_callability_fraction": 0.8,
+    "mutation_embedding_max_norm": 20.0,
+    "nhead": 8,
+    "num_encoder_layers": 3,
+    "optimizer": "AdamW",
+    "optimizer_foreach": null,
+    "population": "sas",
+    "position_encoding": "modulo",
+    "precision": "32-true",
+    "pu_class_prior": null,
+    "pusb": true,
+    "return_attn": true,
+    "scheduler": "CosineAnnealingLR",
+    "seed": 482,
+    "shuffle_pvc_across_genes": false,
+    "stop_on_nan": true,
+    "threshold": 0.6300037906275539,
+    "threshold_positive_fraction": null,
+    "use_gc": true,
+    "use_go": true,
+    "use_pvc": true,
+    "variant_frequency_qc": false,
+    "variant_head": true,
+    "variant_loss_weight": 0.1,
+    "variant_objective": "pooled_bce",
+    "variant_pairs_per_gene": 256,
+    "variant_pairs_per_side": 96,
+    "variant_token_ablate": [],
+    "variant_token_schema": "original",
+    "weight_decay": 0.0003
+  },
+  "parameters": 8378150,
+  "population": "sas",
+  "seed": 482,
+  "selected_epoch": 21,
+  "source_check": "evaluation_manifest",
+  "source_checkpoint_sha256": "5611864f162ff430d278fba27b80e852995b28be4e91843e8c503276f75d7c21",
+  "validation_spearman": 0.41,
+  "variant_head": true,
+  "weights_sha256": "ebf099391138a2d98189cd6fcb7d6117d9d274674f8eeeef18aa5f3907d83f57"
+}

sas/seed482.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ebf099391138a2d98189cd6fcb7d6117d9d274674f8eeeef18aa5f3907d83f57
+size 33521572