\nI have been trying to submit a submission test file with random numbers entered to check the file format is correct, and I keep running into failures. Any of you kind and intelligent souls know why?
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F22687101%2F09759580fddaa729aa98fdb11c6624ab%2FRNA%20Fail.png?generation=1741786189026519&alt=media\")
Hi All
\nI have been trying to submit a submission test file with random numbers entered to check the file format is correct, and I keep running into failures. Any of you kind and intelligent souls know why?
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F22687101%2Fcbb2c0be26a94a5215f7cac5559dcc8d%2FRNA%20Fail2.png?generation=1741786206957989&alt=media\")
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F22687101%2F4b1cae59dbb535ec4655f44b7a6cd948%2FRNAFail3.png?generation=1741786215901988&alt=media\")
Hi All
\nI have been trying to submit a submission test file with random numbers entered to check the file format is correct, and I keep running into failures. Any of you kind and intelligent souls know why?
there's an error symbol (warning triangle) next to your submission, you can click on that warning triangle icon to see more details (although very brief) about what caused the error.
", "rawMarkdown": "there's an error symbol (warning triangle) next to your submission, you can click on that warning triangle icon to see more details (although very brief) about what caused the error.", "votes": null }, { "id": "3148165", "postDate": "03/12/2025 19:32:51", "content": "You can Use My Inference (if works with your Environment)
\n \n all_preds = []\n i tqdm(((test_dataset)), desc=):\n sample = test_dataset[i]\n src = sample[].unsqueeze().to(device)\n mask = sample[].unsqueeze().to(device)\n preds = []\n\n model.train()\n _ ():\n torch.no_grad():\n xyz = model(src, src_mask=mask)\n \n pad_size = (sample[]) - xyz.shape[] \n pad_size > :\n padding = torch.zeros(, pad_size, , device=xyz.device) \n xyz = torch.cat([xyz, padding], dim=) \n preds.append(xyz.squeeze().cpu().numpy())\n\n\n model.()\n torch.no_grad():\n xyz = model(src, src_mask=mask)\n \n pad_size = (sample[]) - xyz.shape[]\n pad_size > :\n padding = torch.zeros(, pad_size, , device=xyz.device)\n xyz = torch.cat([xyz, padding], dim=)\n preds.append(xyz.squeeze().cpu().numpy())\n\n preds = np.stack(preds, axis=)\n all_preds.append(preds)\n\n\n \n submission_data = []\n i, sample (tqdm(test_dataset, desc=)):\n target_id = sample[]\n seq_str = sample[] \n preds = all_preds[i]\n L = (seq_str) \n j (L):\n row = [, seq_str[j], j+]\n k ():\n x, y, z = preds[k][j]\n np.isnan(x) np.isinf(x):\n x = \n np.isnan(y) np.isinf(y):\n y = \n np.isnan(z) np.isinf(z):\n z = \n row.extend([, , ])\n submission_data.append(row)\n\n columns = [, , ]\n i (, ):\n columns.extend([, , ])\n\n submission_df = pd.DataFrame(submission_data, columns=columns)\n\n \n i (, ):\n submission_df[] = submission_df[].astype()\n submission_df[] = submission_df[].astype()\n submission_df[] = submission_df[].astype()\n submission_df[] = submission_df[].astype()\n submission_df = submission_df.fillna()\n submission_df.to_csv(, index=)\n ()\n (submission_df.head())\nMake sure you extract correct logic for correct parameters!
", "rawMarkdown": "You can Use My Inference (if works with your Environment)\n```python\n\n # --- Inference ---\n all_preds = []\n for i in tqdm(range(len(test_dataset)), desc=\"Inference\"):\n sample = test_dataset[i]\n src = sample['sequence'].unsqueeze(0).to(device)\n mask = sample['mask'].unsqueeze(0).to(device)\n preds = []\n\n model.train()\n for _ in range(4):\n with torch.no_grad():\n xyz = model(src, src_mask=mask)\n # Crucially, pad the predictions with zeros to match the *original* sequence length.\n pad_size = len(sample['seq_str']) - xyz.shape[1] # Calculate padding needed\n if pad_size > 0:\n padding = torch.zeros(1, pad_size, 3, device=xyz.device) # Create padding tensor\n xyz = torch.cat([xyz, padding], dim=1) # Concatenate padding\n preds.append(xyz.squeeze(0).cpu().numpy())\n\n\n model.eval()\n with torch.no_grad():\n xyz = model(src, src_mask=mask)\n # Pad the predictions (same as above).\n pad_size = len(sample['seq_str']) - xyz.shape[1]\n if pad_size > 0:\n padding = torch.zeros(1, pad_size, 3, device=xyz.device)\n xyz = torch.cat([xyz, padding], dim=1)\n preds.append(xyz.squeeze(0).cpu().numpy())\n\n preds = np.stack(preds, axis=0)\n all_preds.append(preds)\n\n\n # --- Create Submission File ---\n submission_data = []\n for i, sample in enumerate(tqdm(test_dataset, desc=\"Creating Submission\")):\n target_id = sample['target_id']\n seq_str = sample['seq_str'] # Now this is the *full* sequence.\n preds = all_preds[i]\n L = len(seq_str) # Use the length of the *full* sequence.\n for j in range(L):\n row = [f\"{target_id}_{j+1}\", seq_str[j], j+1]\n for k in range(5):\n x, y, z = preds[k][j]\n if np.isnan(x) or np.isinf(x):\n x = 0.0\n if np.isnan(y) or np.isinf(y):\n y = 0.0\n if np.isnan(z) or np.isinf(z):\n z = 0.0\n row.extend([f\"{x:.3f}\", f\"{y:.3f}\", f\"{z:.3f}\"])\n submission_data.append(row)\n\n columns = ['ID', 'resname', 'resid']\n for i in range(1, 6):\n columns.extend([f\"x_{i}\", f\"y_{i}\", f\"z_{i}\"])\n\n submission_df = pd.DataFrame(submission_data, columns=columns)\n\n # --- Force Data Types (Important) ---\n for i in range(1, 6):\n submission_df[f'x_{i}'] = submission_df[f'x_{i}'].astype(float)\n submission_df[f'y_{i}'] = submission_df[f'y_{i}'].astype(float)\n submission_df[f'z_{i}'] = submission_df[f'z_{i}'].astype(float)\n submission_df['resid'] = submission_df['resid'].astype(int)\n submission_df = submission_df.fillna(0.0)\n submission_df.to_csv(\"submission.csv\", index=False)\n print(\"Submission file created: submission.csv\")\n print(submission_df.head())\n```\n\nMake sure you extract correct logic for correct parameters!", "votes": null } ], "comments": [ { "id": 3147892, "author_name": "jaejohn", "author_url": "", "post_date": "03/12/2025 14:15:09", "content": "there's an error symbol (warning triangle) next to your submission, you can click on that warning triangle icon to see more details (although very brief) about what caused the error.
", "votes": null, "replies": [] }, { "id": 3148165, "author_name": "sangrampatil5150", "author_url": "", "post_date": "03/12/2025 19:32:51", "content": "You can Use My Inference (if works with your Environment)
\n \n all_preds = []\n i tqdm(((test_dataset)), desc=):\n sample = test_dataset[i]\n src = sample[].unsqueeze().to(device)\n mask = sample[].unsqueeze().to(device)\n preds = []\n\n model.train()\n _ ():\n torch.no_grad():\n xyz = model(src, src_mask=mask)\n \n pad_size = (sample[]) - xyz.shape[] \n pad_size > :\n padding = torch.zeros(, pad_size, , device=xyz.device) \n xyz = torch.cat([xyz, padding], dim=) \n preds.append(xyz.squeeze().cpu().numpy())\n\n\n model.()\n torch.no_grad():\n xyz = model(src, src_mask=mask)\n \n pad_size = (sample[]) - xyz.shape[]\n pad_size > :\n padding = torch.zeros(, pad_size, , device=xyz.device)\n xyz = torch.cat([xyz, padding], dim=)\n preds.append(xyz.squeeze().cpu().numpy())\n\n preds = np.stack(preds, axis=)\n all_preds.append(preds)\n\n\n \n submission_data = []\n i, sample (tqdm(test_dataset, desc=)):\n target_id = sample[]\n seq_str = sample[] \n preds = all_preds[i]\n L = (seq_str) \n j (L):\n row = [, seq_str[j], j+]\n k ():\n x, y, z = preds[k][j]\n np.isnan(x) np.isinf(x):\n x = \n np.isnan(y) np.isinf(y):\n y = \n np.isnan(z) np.isinf(z):\n z = \n row.extend([, , ])\n submission_data.append(row)\n\n columns = [, , ]\n i (, ):\n columns.extend([, , ])\n\n submission_df = pd.DataFrame(submission_data, columns=columns)\n\n \n i (, ):\n submission_df[] = submission_df[].astype()\n submission_df[] = submission_df[].astype()\n submission_df[] = submission_df[].astype()\n submission_df[] = submission_df[].astype()\n submission_df = submission_df.fillna()\n submission_df.to_csv(, index=)\n ()\n (submission_df.head())\nMake sure you extract correct logic for correct parameters!
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "3147848": "Hi All\nI have been trying to submit a submission test file with random numbers entered to check the file format is correct, and I keep running into failures. Any of you kind and intelligent souls know why?\n\n\n\n", "3147892": "there's an error symbol (warning triangle) next to your submission, you can click on that warning triangle icon to see more details (although very brief) about what caused the error.", "3148165": "You can Use My Inference (if works with your Environment)\n```python\n\n # --- Inference ---\n all_preds = []\n for i in tqdm(range(len(test_dataset)), desc=\"Inference\"):\n sample = test_dataset[i]\n src = sample['sequence'].unsqueeze(0).to(device)\n mask = sample['mask'].unsqueeze(0).to(device)\n preds = []\n\n model.train()\n for _ in range(4):\n with torch.no_grad():\n xyz = model(src, src_mask=mask)\n # Crucially, pad the predictions with zeros to match the *original* sequence length.\n pad_size = len(sample['seq_str']) - xyz.shape[1] # Calculate padding needed\n if pad_size > 0:\n padding = torch.zeros(1, pad_size, 3, device=xyz.device) # Create padding tensor\n xyz = torch.cat([xyz, padding], dim=1) # Concatenate padding\n preds.append(xyz.squeeze(0).cpu().numpy())\n\n\n model.eval()\n with torch.no_grad():\n xyz = model(src, src_mask=mask)\n # Pad the predictions (same as above).\n pad_size = len(sample['seq_str']) - xyz.shape[1]\n if pad_size > 0:\n padding = torch.zeros(1, pad_size, 3, device=xyz.device)\n xyz = torch.cat([xyz, padding], dim=1)\n preds.append(xyz.squeeze(0).cpu().numpy())\n\n preds = np.stack(preds, axis=0)\n all_preds.append(preds)\n\n\n # --- Create Submission File ---\n submission_data = []\n for i, sample in enumerate(tqdm(test_dataset, desc=\"Creating Submission\")):\n target_id = sample['target_id']\n seq_str = sample['seq_str'] # Now this is the *full* sequence.\n preds = all_preds[i]\n L = len(seq_str) # Use the length of the *full* sequence.\n for j in range(L):\n row = [f\"{target_id}_{j+1}\", seq_str[j], j+1]\n for k in range(5):\n x, y, z = preds[k][j]\n if np.isnan(x) or np.isinf(x):\n x = 0.0\n if np.isnan(y) or np.isinf(y):\n y = 0.0\n if np.isnan(z) or np.isinf(z):\n z = 0.0\n row.extend([f\"{x:.3f}\", f\"{y:.3f}\", f\"{z:.3f}\"])\n submission_data.append(row)\n\n columns = ['ID', 'resname', 'resid']\n for i in range(1, 6):\n columns.extend([f\"x_{i}\", f\"y_{i}\", f\"z_{i}\"])\n\n submission_df = pd.DataFrame(submission_data, columns=columns)\n\n # --- Force Data Types (Important) ---\n for i in range(1, 6):\n submission_df[f'x_{i}'] = submission_df[f'x_{i}'].astype(float)\n submission_df[f'y_{i}'] = submission_df[f'y_{i}'].astype(float)\n submission_df[f'z_{i}'] = submission_df[f'z_{i}'].astype(float)\n submission_df['resid'] = submission_df['resid'].astype(int)\n submission_df = submission_df.fillna(0.0)\n submission_df.to_csv(\"submission.csv\", index=False)\n print(\"Submission file created: submission.csv\")\n print(submission_df.head())\n```\n\nMake sure you extract correct logic for correct parameters!" }, "source": "meta" }