Remove LFS cache

unconditioned_geom.yaml

@@ -1,43 +0,0 @@
-actual_steps: 19
-score_ckpt: resources/checkpoints/DiffDock/score_model/best_ema_inference_epoch_model.pt
-confidence_ckpt: resources/checkpoints/DiffDock/confidence_model/best_model_epoch75.pt
-#confidence_model_dir: ./workdir/v1.1/confidence_model
-different_schedules: false
-inf_sched_alpha: 1
-inf_sched_beta: 1
-inference_steps: 20
-initial_noise_std_proportion: 1.4601642460337794
-limit_failures: 5
-#model_dir: ./workdir/v1.1/score_model
-#comment
-no_final_step_noise: true
-no_model: false
-no_random: false
-no_random_pocket: false
-ode: false
-old_filtering_model: true
-old_score_model: false
-resample_rdkit: false
-samples_per_complex: 10
-sigma_schedule: expbeta
-temp_psi_rot: 0.9022615585677628
-temp_psi_tor: 0.5946212391366862
-temp_psi_tr: 0.727287304570729
-temp_sampling_rot: 2.06391612594481
-temp_sampling_tor: 7.044261621607846
-temp_sampling_tr: 1.170050527854316
-temp_sigma_data_rot: 0.7464326999906034
-temp_sigma_data_tor: 0.6943254174849822
-temp_sigma_data_tr: 0.9299802531572672
-
-eps: 5
-min_samples: 10
-frag_dist_range: (2, 5)
-distance_type: min
-
-linker_ckpt:
-  pocket_full: resources/checkpoints/DiffLinker/pockets_difflinker_full_no_anchors_fc_pdb_excluded.ckpt
-  pocket_bb: resources/checkpoints/DiffLinker/pockets_difflinker_backbone.ckpt
-  geom: resources/checkpoints/DiffLinker/geom_difflinker.ckpt
-linker_size: resources/checkpoints/DiffLinker/geom_size_gnn.ckpt
-linker_condition: 'protein'  # pocket

utils/so3.py

@@ -42,11 +42,11 @@ def _score(exp, omega, eps, L=2000):  # score of density over SO(3)
     return dSigma / exp
 
 
-if os.path.exists('.so3_omegas_array4.npy'):
-    _omegas_array = np.load('.so3_omegas_array4.npy')
-    _cdf_vals = np.load('.so3_cdf_vals4.npy')
-    _score_norms = np.load('.so3_score_norms4.npy')
-    _exp_score_norms = np.load('.so3_exp_score_norms4.npy')
+if os.path.exists('resources/precomputed/.so3_omegas_array4.npy'):
+    _omegas_array = np.load('resources/precomputed/.so3_omegas_array4.npy')
+    _cdf_vals = np.load('resources/precomputed/.so3_cdf_vals4.npy')
+    _score_norms = np.load('resources/precomputed/.so3_score_norms4.npy')
+    _exp_score_norms = np.load('resources/precomputed/.so3_exp_score_norms4.npy')
 else:
     _eps_array = 10 ** np.linspace(np.log10(MIN_EPS), np.log10(MAX_EPS), N_EPS)
     _omegas_array = np.linspace(0, np.pi, X_N + 1)[1:]
@@ -58,10 +58,10 @@ else:
 
     _exp_score_norms = np.sqrt(np.sum(_score_norms**2 * _pdf_vals, axis=1) / np.sum(_pdf_vals, axis=1) / np.pi)
 
-    np.save('.so3_omegas_array4.npy', _omegas_array)
-    np.save('.so3_cdf_vals4.npy', _cdf_vals)
-    np.save('.so3_score_norms4.npy', _score_norms)
-    np.save('.so3_exp_score_norms4.npy', _exp_score_norms)
+    np.save('resources/precomputed/.so3_omegas_array4.npy', _omegas_array)
+    np.save('resources/precomputed/.so3_cdf_vals4.npy', _cdf_vals)
+    np.save('resources/precomputed/.so3_score_norms4.npy', _score_norms)
+    np.save('resources/precomputed/.so3_exp_score_norms4.npy', _exp_score_norms)
 
 
 def sample(eps):

utils/torus.py

@@ -28,12 +28,12 @@ SIGMA_MIN, SIGMA_MAX, SIGMA_N = 3e-3, 2, 5000  # relative to pi
 x = 10 ** np.linspace(np.log10(X_MIN), 0, X_N + 1) * np.pi
 sigma = 10 ** np.linspace(np.log10(SIGMA_MIN), np.log10(SIGMA_MAX), SIGMA_N + 1) * np.pi
 
-if os.path.exists('.p.npy'):
-    p_ = np.load('.p.npy')
-    score_ = np.load('.score.npy')
+if os.path.exists('resources/precomputed/.p.npy'):
+    p_ = np.load('resources/precomputed/.p.npy')
+    score_ = np.load('resources/precomputed/.score.npy')
 else:
     p_ = p(x, sigma[:, None], N=100)
-    np.save('.p.npy', p_)
+    np.save('resources/precomputed/.p.npy', p_)
 
     eps = np.finfo(p_.dtype).eps
     score_ = grad(x, sigma[:, None], N=100) / (p_ + eps)