Add FIM-PP paper checkpoints (Table 1 fine-tuned, Table 2 zero-shot and fine-tuned)
d642f39 verified | { | |
| "alpha_decoder": { | |
| "hidden_act": { | |
| "name": "torch.nn.GELU" | |
| }, | |
| "hidden_layers": [ | |
| 256, | |
| 256 | |
| ], | |
| "name": "fim.models.blocks.base.MLP" | |
| }, | |
| "architectures": [ | |
| "FIMHawkes" | |
| ], | |
| "beta_decoder": { | |
| "hidden_act": { | |
| "name": "torch.nn.GELU" | |
| }, | |
| "hidden_layers": [ | |
| 256, | |
| 256 | |
| ], | |
| "name": "fim.models.blocks.base.MLP" | |
| }, | |
| "context_summary_encoder": { | |
| "encoder_layer": { | |
| "batch_first": true, | |
| "dropout": 0.0, | |
| "name": "torch.nn.TransformerEncoderLayer", | |
| "nhead": 4 | |
| }, | |
| "name": "torch.nn.TransformerEncoder", | |
| "num_layers": 2 | |
| }, | |
| "context_summary_pooling": { | |
| "attention": { | |
| "nhead": 4 | |
| }, | |
| "name": "fim.models.blocks.neural_operators.AttentionOperator", | |
| "num_res_layers": 1, | |
| "paths_block_attention": false | |
| }, | |
| "context_ts_encoder": { | |
| "encoder_layer": { | |
| "batch_first": true, | |
| "dropout": 0.0, | |
| "name": "torch.nn.TransformerEncoderLayer", | |
| "nhead": 4 | |
| }, | |
| "name": "torch.nn.TransformerEncoder", | |
| "num_layers": 4 | |
| }, | |
| "decoder_ts": { | |
| "decoder_layer": { | |
| "batch_first": true, | |
| "dropout": 0.0, | |
| "name": "torch.nn.TransformerDecoderLayer", | |
| "nhead": 4 | |
| }, | |
| "name": "torch.nn.TransformerDecoder", | |
| "num_layers": 4 | |
| }, | |
| "delta_time_encoder": { | |
| "name": "fim.models.blocks.positional_encodings.SineTimeEncoding", | |
| "out_features": 256 | |
| }, | |
| "evaluation_mark_encoder": { | |
| "name": "torch.nn.Linear" | |
| }, | |
| "hidden_act": { | |
| "name": "torch.nn.GELU" | |
| }, | |
| "hidden_dim": 256, | |
| "loss_weights": { | |
| "alpha": 0.0, | |
| "mu": 0.0, | |
| "nll": 1.0, | |
| "relative_spike": 0.0, | |
| "smape": 0.0 | |
| }, | |
| "mark_encoder": { | |
| "name": "torch.nn.Linear", | |
| "out_features": 256 | |
| }, | |
| "mark_fusion_attention": null, | |
| "max_num_marks": 22, | |
| "mu_decoder": { | |
| "hidden_act": { | |
| "name": "torch.nn.GELU" | |
| }, | |
| "hidden_layers": [ | |
| 256, | |
| 256 | |
| ], | |
| "name": "fim.models.blocks.base.MLP" | |
| }, | |
| "nll": { | |
| "method": "monte_carlo", | |
| "num_integration_points": 200 | |
| }, | |
| "normalize_by_max_time": false, | |
| "normalize_times": true, | |
| "thinning": null, | |
| "time_encoder": { | |
| "name": "fim.models.blocks.positional_encodings.SineTimeEncoding", | |
| "out_features": 256 | |
| }, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.46.0", | |
| "model_type": "fimhawkes" | |
| } | |