[WIP] remove redundant / unused code #3
Draft
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
linziyi96
changed the title
linziyi96
added a commit
that referenced
this pull request
Aug 2, 2023
It is probably safer to keep CLIP at its original precision (e.g., fp16) regardless of the autocast setting: Some casting (e.g., from fp16 to bf16) may be lossy and can potentially harm the pre-trained model. Keep the changes to llama.py only at this moment since a lot of copy- pasted codes may be refactored in the future (#3).
linziyi96
added a commit
that referenced
this pull request
Aug 4, 2023
…ze. (#16) * temp save * quick fix of demo memory issue * Refactor tensor creation dtype / device control. This commit makes two changes during model creation: 1. Decouples promote_trainable_params_to_fp32 from model __init__. This is to avoid casting to fp32 to save memory in inference-only mode (#4). 2. Use a context manager to manage default tensor type change. In the previous version, the default tensor type is reset to torch.FloatTensor after creating the vision model, which is technically incorrect and should be the previous default tensor type instead. We implement our own context manager because the official context managers seem to be incomplete at this time (PyTorch 2.0.1): No dtype manager is provided and set_default_device is ineffective to the torch.Tensor calls which are used in fairscale. * Change CLIP dtype management in llama.py It is probably safer to keep CLIP at its original precision (e.g., fp16) regardless of the autocast setting: Some casting (e.g., from fp16 to bf16) may be lossy and can potentially harm the pre-trained model. Keep the changes to llama.py only at this moment since a lot of copy- pasted codes may be refactored in the future (#3). * Respect args.precision when saving checkpoints. * Support checkpoint merge Checkpoint merge is suported in misc/tensor_parallel.py. Merge requires that the checkpoint_mp_world_size % mp_world_size == 0. Support for split (i.e., when mp_world_size % checkpoint_mp_world_size == 0) and redistribute (for general mp_world_size and checkpoint_mp_world_size values) will be added in the future. Also changing multi_turn demo to use the new loading function with merge support. * move printing trainable params * move training model creation back to cpu Closes #15, #13
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR marks a series of changes to replace copy-pasted code with import / inheritance to improve maintainability. The update plan and progress is listed as follows. It is NOT ready to be merged at this moment.