params.yaml

# FLARE — FILIP contrastive retrieval (SpecFormula + GNN).
# Paths below are relative to the repository root unless absolute.
# Override file location: FLARE_PARAMS=/path/to/custom.yaml

# --- Run / logging ---
run_name: flare_filip
job_key: ''
run_details: ''
project_name: ''
wandb_entity_name: ''
no_wandb: true
seed: 0
debug: false
bootstrap_metrics: true
log_only_loss_at_stages: []

# --- Training ---
lr: 2.881339661302105e-05
weight_decay: 1.1586679936312845e-05
max_epochs: 2000
accelerator: gpu
devices: [1]
batch_size: 64
num_workers: 50
log_every_n_steps: 250
val_check_interval: 1.0
early_stopping_patience: 300
contr_temp: 0.022772534845886608
loss_strategy: static
checkpoint_pth: null

# --- Data (set to your local copies) ---
dataset_pth: data/MassSpecGym.tsv
candidates_pth: data/MassSpecGym_retrieval_candidates_mass.json
subformula_dir_pth: data/subformulae_default
split_pth: null
formula_source: default

# --- Views / featurizers ---
spectra_view: SpecFormula
molecule_view: MolGraph
max_mz: 1000
bin_width: 1
mask_peak_ratio: 0.0
mask_precursor: false
element_list:
  - H
  - C
  - O
  - N
  - P
  - S
  - Cl
  - F
  - Br
  - I
  - B
  - As
  - Si
  - Se
add_intensities: true
atom_feature: full
bond_feature: full

# --- Task & model ---
task: retrieval
spec_enc: Transformer_Formula
mol_enc: GNN
model: filipContrastive
contr_views:
  - [spec_enc, mol_enc]

final_embedding_dim: 512
fc_dropout: 0.4
hidden_dims: [64, 128]

formula_dropout: 0.2
formula_dims: [512, 256, 512]
cross_attn_heads: 2
use_cls: false
peak_dropout: 0.414425691950033
formula_attn_heads: 4
formula_transformer_layers: 2
attn_heads: [12, 12, 12]

gnn_channels: [64, 128, 512]
gnn_type: gcn
num_gnn_layers: 3
gnn_hidden_dim: 512
gnn_dropout: 0.23234950970370824

# --- Evaluation (retrieval base class) ---
at_ks: [1, 5, 20]
myopic_mces_kwargs: null

# --- Outputs (optional; train/test override as needed) ---
df_test_path: ''