sweep-ai[bot]
commented
1 year ago ❌ Unable to Complete PR
I'm sorry, but it looks like an error has occurred. Try changing the issue description to re-trigger Sweep. If this error persists contact team@sweep.dev.
For bonus GPT-4 tickets, please report this bug on Discord.
Please look at the generated plan. If something looks wrong, please add more details to your issue.
| File Path | Proposed Changes |
|---|---|
deltalm/fairseq/fairseq/checkpoint_utils.py |
Create deltalm/fairseq/fairseq/checkpoint_utils.py with contents: • Locate the '_upgrade_state_dict' function. • Find the line where 'best_loss' is accessed from the state dictionary. • Before this line, add a check to see if 'best_loss' is in the state dictionary. If it is not, add it to the state dictionary with a default value of None. |
To recreate the pull request edit the issue title or description.
kevinlu1248
Khaled-Elsaka
in the file unilm/deltalm/test.py i am trying to load deltalm model but I got this error, I think that the model.pt itself doesn't have the required configuration or model state can you help me fix this.
KeyError Traceback (most recent call last)
4 frames /content/drive/MyDrive/Translation Fine-Tuning/Exploring new models/unilm/deltalm/fairseq/fairseq/models/fairseq_model.py in from_pretrained(cls, model_name_or_path, checkpoint_file, data_name_or_path, **kwargs) 265 from fairseq import hub_utils 266 --> 267 x = hub_utils.from_pretrained( 268 model_name_or_path, 269 checkpoint_file,
/content/drive/MyDrive/Translation Fine-Tuning/Exploring new models/unilm/deltalm/fairseq/fairseq/hub_utils.py in from_pretrained(model_name_or_path, checkpoint_file, data_name_or_path, archive_map, **kwargs) 71 utils.import_user_module(argparse.Namespace(user_dir=kwargs["user_dir"])) 72 ---> 73 models, args, task = checkpoint_utils.load_model_ensemble_and_task( 74 [os.path.join(model_path, cpt) for cpt in checkpoint_file.split(os.pathsep)], 75 arg_overrides=kwargs,
/content/drive/MyDrive/Translation Fine-Tuning/Exploring new models/unilm/deltalm/fairseq/fairseq/checkpoint_utils.py in load_model_ensemble_and_task(filenames, arg_overrides, task, strict, suffix, num_shards, state) 419 raise IOError("Model file not found: {}".format(filename)) 420 if state is None: --> 421 state = load_checkpoint_to_cpu(filename, arg_overrides) 422 if "args" in state and state["args"] is not None: 423 cfg = convert_namespace_to_omegaconf(state["args"])
/content/drive/MyDrive/Translation Fine-Tuning/Exploring new models/unilm/deltalm/fairseq/fairseq/checkpoint_utils.py in load_checkpoint_to_cpu(path, arg_overrides, load_on_all_ranks) 337 overwrite_args_by_name(state["cfg"], arg_overrides) 338 --> 339 state = _upgrade_state_dict(state) 340 return state 341
/content/drive/MyDrive/Translation Fine-Tuning/Exploring new models/unilm/deltalm/fairseq/fairseq/checkpoint_utils.py in _upgrade_state_dict(state) 579 if "optimizer_history" not in state: 580 state["optimizer_history"] = [ --> 581 {"criterion_name": "CrossEntropyCriterion", "best_loss": state["best_loss"]} 582 ] 583 state["last_optimizer_state"] = state["optimizer"]
KeyError: 'best_loss'
Checklist
- [ ] `deltalm/fairseq/fairseq/checkpoint_utils.py` > • Locate the '_upgrade_state_dict' function. > • Find the line where 'best_loss' is accessed from the state dictionary. > • Before this line, add a check to see if 'best_loss' is in the state dictionary. If it is not, add it to the state dictionary with a default value of None.