Ensure additional metadata to trackers don't throw error in happy case.

[dushyantbehl]

Description of the change

Ensure additional metadata to trackers will not throw errors in happy case.

Related issue number

Fixes #289

How to verify the PR

Verified

(.fms-hf-tuning-venv) ➜  fms-hf-tuning git:(fix-bad-value-error) ✗ accelerate launch \
--main_process_port 1234 \
--config_file /home/schrodinger/workspace/ibm/fms-hf-tuning/fixtures/accelerate_fsdp_defaults.yaml \
--num_processes 2 \
-m tuning.sft_trainer \
--model_name_or_path "Maykeye/TinyLLama-v0" \
--training_data_path "/home/schrodinger/bin/twitter_complaints.json" \
--output_dir /home/schrodinger/bin/output/tiny-llama-ft-multigpu \
--num_train_epochs 5 \
--per_device_train_batch_size 4 \
--gradient_accumulation_steps 4 \
--per_device_eval_batch_size 4 \
--learning_rate 0.000001 \
--use_flash_attn no \
--response_template "\n### Label:" \
--dataset_text_field "output" \
--tracker aim \
--aim_repo /home/schrodinger/bin/aimrepo/ \
--experiment aim-test-distributed-main \
--torch_dtype float16 
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/accelerate/utils/launch.py:253: FutureWarning: `fsdp_backward_prefetch_policy` is deprecated and will be removed in version 0.27.0 of 🤗 Accelerate. Use `fsdp_backward_prefetch` instead
  warnings.warn(
You are using the default legacy behaviour of the <class 'transformers.models.llama.tokenization_llama_fast.LlamaTokenizerFast'>. This is expected, and simply means that the `legacy` (previous) behavior will be used so nothing changes for you. If you want to use the new behaviour, set `legacy=False`. This should only be set if you understand what it means, and thoroughly read the reason why this was added as explained in https://github.com/huggingface/transformers/pull/24565 - if you loaded a llama tokenizer from a GGUF file you can ignore this message.
You are using the default legacy behaviour of the <class 'transformers.models.llama.tokenization_llama_fast.LlamaTokenizerFast'>. This is expected, and simply means that the `legacy` (previous) behavior will be used so nothing changes for you. If you want to use the new behaviour, set `legacy=False`. This should only be set if you understand what it means, and thoroughly read the reason why this was added as explained in https://github.com/huggingface/transformers/pull/24565 - if you loaded a llama tokenizer from a GGUF file you can ignore this message.
max_seq_length 4096 exceeds tokenizer.model_max_length             2048, using tokenizer.model_max_length 2048
max_seq_length 4096 exceeds tokenizer.model_max_length             2048, using tokenizer.model_max_length 2048
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/transformers/training_args.py:2007: FutureWarning: `--push_to_hub_token` is deprecated and will be removed in version 5 of 🤗 Transformers. Use `--hub_token` instead.
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/huggingface_hub/utils/_deprecation.py:100: FutureWarning: Deprecated argument(s) used in '__init__': dataset_text_field, max_seq_length. Will not be supported from version '1.0.0'.

Deprecated positional argument(s) used in SFTTrainer, please use the SFTConfig to set these arguments instead.
  warnings.warn(message, FutureWarning)
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/trl/trainer/sft_trainer.py:280: UserWarning: You passed a `max_seq_length` argument to the SFTTrainer, the value you passed will override the one in the `SFTConfig`.
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/trl/trainer/sft_trainer.py:318: UserWarning: You passed a `dataset_text_field` argument to the SFTTrainer, the value you passed will override the one in the `SFTConfig`.
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/transformers/training_args.py:2007: FutureWarning: `--push_to_hub_token` is deprecated and will be removed in version 5 of 🤗 Transformers. Use `--hub_token` instead.
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/huggingface_hub/utils/_deprecation.py:100: FutureWarning: Deprecated argument(s) used in '__init__': dataset_text_field, max_seq_length. Will not be supported from version '1.0.0'.

Deprecated positional argument(s) used in SFTTrainer, please use the SFTConfig to set these arguments instead.
  warnings.warn(message, FutureWarning)
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/trl/trainer/sft_trainer.py:280: UserWarning: You passed a `max_seq_length` argument to the SFTTrainer, the value you passed will override the one in the `SFTConfig`.
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/trl/trainer/sft_trainer.py:318: UserWarning: You passed a `dataset_text_field` argument to the SFTTrainer, the value you passed will override the one in the `SFTConfig`.
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/trl/trainer/sft_trainer.py:408: UserWarning: You passed a tokenizer with `padding_side` not equal to `right` to the SFTTrainer. This might lead to some unexpected behaviour due to overflow issues when training a model in half-precision. You might consider adding `tokenizer.padding_side = 'right'` to your code.
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/trl/trainer/sft_trainer.py:408: UserWarning: You passed a tokenizer with `padding_side` not equal to `right` to the SFTTrainer. This might lead to some unexpected behaviour due to overflow issues when training a model in half-precision. You might consider adding `tokenizer.padding_side = 'right'` to your code.
  warnings.warn(
{'loss': 22276357.5138, 'grad_norm': nan, 'learning_rate': 8e-07, 'epoch': 1.0}                                                                                                                                                               
 20%|███████████████████████████████████████▌                                                                                                                                                              | 325/1625 [01:54<07:28,  2.90it/s]/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py:689: FutureWarning: FSDP.state_dict_type() and FSDP.set_state_dict_type() are being deprecated. Please use APIs, get_state_dict() and set_state_dict(), which can support different parallelisms, FSDP1, FSDP2, DDP. API doc: https://pytorch.org/docs/stable/distributed.checkpoint.html#torch.distributed.checkpoint.state_dict.get_state_dict .Tutorial: https://pytorch.org/tutorials/recipes/distributed_checkpoint_recipe.html .
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py:689: FutureWarning: FSDP.state_dict_type() and FSDP.set_state_dict_type() are being deprecated. Please use APIs, get_state_dict() and set_state_dict(), which can support different parallelisms, FSDP1, FSDP2, DDP. API doc: https://pytorch.org/docs/stable/distributed.checkpoint.html#torch.distributed.checkpoint.state_dict.get_state_dict .Tutorial: https://pytorch.org/tutorials/recipes/distributed_checkpoint_recipe.html .
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py:689: FutureWarning: FSDP.state_dict_type() and FSDP.set_state_dict_type() are being deprecated. Please use APIs, get_state_dict() and set_state_dict(), which can support different parallelisms, FSDP1, FSDP2, DDP. API doc: https://pytorch.org/docs/stable/distributed.checkpoint.html#torch.distributed.checkpoint.state_dict.get_state_dict .Tutorial: https://pytorch.org/tutorials/recipes/distributed_checkpoint_recipe.html .
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/storage.py:414: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
  return torch.load(io.BytesIO(b))
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/storage.py:414: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
  return torch.load(io.BytesIO(b))
{'loss': 0.0, 'grad_norm': nan, 'learning_rate': 6e-07, 'epoch': 2.0}                                                                                                                                                                         
 40%|███████████████████████████████████████████████████████████████████████████████▏                                                                                                                      | 650/1625 [03:48<06:13,  2.61it/s]/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py:689: FutureWarning: FSDP.state_dict_type() and FSDP.set_state_dict_type() are being deprecated. Please use APIs, get_state_dict() and set_state_dict(), which can support different parallelisms, FSDP1, FSDP2, DDP. API doc: https://pytorch.org/docs/stable/distributed.checkpoint.html#torch.distributed.checkpoint.state_dict.get_state_dict .Tutorial: https://pytorch.org/tutorials/recipes/distributed_checkpoint_recipe.html .
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py:689: FutureWarning: FSDP.state_dict_type() and FSDP.set_state_dict_type() are being deprecated. Please use APIs, get_state_dict() and set_state_dict(), which can support different parallelisms, FSDP1, FSDP2, DDP. API doc: https://pytorch.org/docs/stable/distributed.checkpoint.html#torch.distributed.checkpoint.state_dict.get_state_dict .Tutorial: https://pytorch.org/tutorials/recipes/distributed_checkpoint_recipe.html .
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/storage.py:414: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
  return torch.load(io.BytesIO(b))
{'loss': 0.0, 'grad_norm': nan, 'learning_rate': 4e-07, 'epoch': 3.0}                                                                                                                                                                         
 60%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████▊                                                                               | 975/1625 [05:42<03:29,  3.10it/s]/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py:689: FutureWarning: FSDP.state_dict_type() and FSDP.set_state_dict_type() are being deprecated. Please use APIs, get_state_dict() and set_state_dict(), which can support different parallelisms, FSDP1, FSDP2, DDP. API doc: https://pytorch.org/docs/stable/distributed.checkpoint.html#torch.distributed.checkpoint.state_dict.get_state_dict .Tutorial: https://pytorch.org/tutorials/recipes/distributed_checkpoint_recipe.html .
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py:689: FutureWarning: FSDP.state_dict_type() and FSDP.set_state_dict_type() are being deprecated. Please use APIs, get_state_dict() and set_state_dict(), which can support different parallelisms, FSDP1, FSDP2, DDP. API doc: https://pytorch.org/docs/stable/distributed.checkpoint.html#torch.distributed.checkpoint.state_dict.get_state_dict .Tutorial: https://pytorch.org/tutorials/recipes/distributed_checkpoint_recipe.html .
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/storage.py:414: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
  return torch.load(io.BytesIO(b))
{'loss': 0.0, 'grad_norm': nan, 'learning_rate': 2e-07, 'epoch': 4.0}                                                                                                                                                                         
 80%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████▌                                       | 1300/1625 [07:34<01:50,  2.94it/s]/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py:689: FutureWarning: FSDP.state_dict_type() and FSDP.set_state_dict_type() are being deprecated. Please use APIs, get_state_dict() and set_state_dict(), which can support different parallelisms, FSDP1, FSDP2, DDP. API doc: https://pytorch.org/docs/stable/distributed.checkpoint.html#torch.distributed.checkpoint.state_dict.get_state_dict .Tutorial: https://pytorch.org/tutorials/recipes/distributed_checkpoint_recipe.html .
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py:689: FutureWarning: FSDP.state_dict_type() and FSDP.set_state_dict_type() are being deprecated. Please use APIs, get_state_dict() and set_state_dict(), which can support different parallelisms, FSDP1, FSDP2, DDP. API doc: https://pytorch.org/docs/stable/distributed.checkpoint.html#torch.distributed.checkpoint.state_dict.get_state_dict .Tutorial: https://pytorch.org/tutorials/recipes/distributed_checkpoint_recipe.html .
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/storage.py:414: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
  return torch.load(io.BytesIO(b))
100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1625/1625 [09:28<00:00,  2.94it/s]/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py:689: FutureWarning: FSDP.state_dict_type() and FSDP.set_state_dict_type() are being deprecated. Please use APIs, get_state_dict() and set_state_dict(), which can support different parallelisms, FSDP1, FSDP2, DDP. API doc: https://pytorch.org/docs/stable/distributed.checkpoint.html#torch.distributed.checkpoint.state_dict.get_state_dict .Tutorial: https://pytorch.org/tutorials/recipes/distributed_checkpoint_recipe.html .
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py:689: FutureWarning: FSDP.state_dict_type() and FSDP.set_state_dict_type() are being deprecated. Please use APIs, get_state_dict() and set_state_dict(), which can support different parallelisms, FSDP1, FSDP2, DDP. API doc: https://pytorch.org/docs/stable/distributed.checkpoint.html#torch.distributed.checkpoint.state_dict.get_state_dict .Tutorial: https://pytorch.org/tutorials/recipes/distributed_checkpoint_recipe.html .
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/storage.py:414: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
  return torch.load(io.BytesIO(b))
{'loss': 0.0, 'grad_norm': nan, 'learning_rate': 0.0, 'epoch': 5.0}                                                                                                                                                                           
100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1625/1625 [09:28<00:00,  2.94it/s]/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py:689: FutureWarning: FSDP.state_dict_type() and FSDP.set_state_dict_type() are being deprecated. Please use APIs, get_state_dict() and set_state_dict(), which can support different parallelisms, FSDP1, FSDP2, DDP. API doc: https://pytorch.org/docs/stable/distributed.checkpoint.html#torch.distributed.checkpoint.state_dict.get_state_dict .Tutorial: https://pytorch.org/tutorials/recipes/distributed_checkpoint_recipe.html .
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py:689: FutureWarning: FSDP.state_dict_type() and FSDP.set_state_dict_type() are being deprecated. Please use APIs, get_state_dict() and set_state_dict(), which can support different parallelisms, FSDP1, FSDP2, DDP. API doc: https://pytorch.org/docs/stable/distributed.checkpoint.html#torch.distributed.checkpoint.state_dict.get_state_dict .Tutorial: https://pytorch.org/tutorials/recipes/distributed_checkpoint_recipe.html .
  warnings.warn(
/home/schrodinger/.fms-hf-tuning-venv/lib/python3.10/site-packages/torch/storage.py:414: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
  return torch.load(io.BytesIO(b))
{'train_runtime': 568.8356, 'train_samples_per_second': 91.415, 'train_steps_per_second': 2.857, 'train_loss': 4455271.502769231, 'epoch': 5.0}                                                                                               
100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1625/1625 [09:28<00:00,  2.86it/s]

Was the PR tested

• I have added >=1 unit test(s) for every new method I have added.
• [x ] I have ensured all unit tests pass

[ashokponkumar]

@kmehant PTAL