Pass quantization_kwargs to CLIP builders #1994

joecummings · 2024-11-12T20:13:08Z

Context

What is the purpose of this PR? Is it to

add a new feature
fix a bug
update tests and/or documentation
other (please add here)

Please link to any issues this PR addresses.

Changelog

What are the changes made in this PR?
*

Test plan

Please make sure to do each of the following if applicable to your PR. If you're unsure about any one of these just ask and we will happily help. We also have a contributing page for some guidance on contributing.

run pre-commit hooks and linters (make sure you've first installed via pre-commit install)
add unit tests for any new functionality
update docstrings for any new or updated methods or classes
run unit tests via pytest tests
run recipe tests via pytest tests -m integration_test
manually run any new or modified recipes with sufficient proof of correctness
include relevant commands and any other artifacts in this summary (pastes of loss curves, eval results, etc.)

(joe-torchtune) [[email protected] ~/projects/joe-torchtune (update-clip-with-quant-nums)]$ tune run --nproc-per-node 8 lora_finetune_distributed --config llama3_2_vision/90B_qlora max_steps_per_epoch=5 lr_scheduler.num_warmup_steps=0
Running with torchrun...
W1112 12:28:55.300000 2440397 site-packages/torch/distributed/run.py:793]
W1112 12:28:55.300000 2440397 site-packages/torch/distributed/run.py:793] *****************************************
W1112 12:28:55.300000 2440397 site-packages/torch/distributed/run.py:793] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed.
W1112 12:28:55.300000 2440397 site-packages/torch/distributed/run.py:793] *****************************************

  _component_: torchtune.modules.loss.CEWithChunkedOutputLoss
lr_scheduler:
  _component_: torchtune.training.lr_schedulers.get_cosine_schedule_with_warmup
  num_warmup_steps: 0
max_steps_per_epoch: 5
metric_logger:
  _component_: torchtune.training.metric_logging.DiskLogger
  log_dir: /tmp/Llama-3.2-90B-Vision-Instruct/logs
model:
  _component_: torchtune.models.llama3_2_vision.qlora_llama3_2_vision_90b
  apply_lora_to_mlp: true
  apply_lora_to_output: false
  decoder_trainable: frozen
  encoder_trainable: lora
  fusion_trainable: lora
  image_size: 560
  lora_alpha: 16
  lora_attn_modules:
  - q_proj
  - v_proj
  - output_proj
  lora_dropout: 0.0
  lora_rank: 8
optimizer:
  _component_: torch.optim.AdamW
  fused: true
  lr: 0.0001
  weight_decay: 0.01
output_dir: /tmp/qlora-llama3.2-vision-finetune
profiler:
  _component_: torchtune.training.setup_torch_profiler
  active_steps: 2
  cpu: true
  cuda: true
  enabled: false
  num_cycles: 1
  output_dir: /tmp/qlora-llama3.2-vision-finetune/profiling_outputs
  profile_memory: false
  record_shapes: true
  wait_steps: 5
  warmup_steps: 3
  with_flops: false
  with_stack: false
resume_from_checkpoint: false
save_adapter_weights_only: false
seed: null
shuffle: true
tokenizer:
  _component_: torchtune.models.llama3_2_vision.llama3_2_vision_transform
  image_size: 560
  max_seq_len: 8192
  path: /tmp/Llama-3.2-90B-Vision-Instruct/original/tokenizer.model

NCCL version 2.21.5+cuda12.4
DEBUG:torchtune.utils._logging:Setting manual seed to local seed 3509117360. Local seed is seed + rank = 3509117360 + 0
Writing logs to /tmp/Llama-3.2-90B-Vision-Instruct/logs/log_1731443356.txt
INFO:torchtune.utils._logging:FSDP is enabled. Instantiating model and loading checkpoint on Rank 0 ...
INFO:torchtune.utils._logging:Instantiating model and loading checkpoint took 127.72 secs
INFO:torchtune.utils._logging:Memory stats after model init:
        GPU peak memory allocation: 7.77 GiB
        GPU peak memory reserved: 9.25 GiB
        GPU peak memory active: 7.77 GiB
INFO:torchtune.utils._logging:Optimizer is initialized.
INFO:torchtune.utils._logging:Loss is initialized.
INFO:torchtune.utils._logging:Dataset and Sampler are initialized.
INFO:torchtune.utils._logging:Learning rate scheduler is initialized.
WARNING:torchtune.utils._logging: Profiling disabled.
INFO:torchtune.utils._logging: Profiler config after instantiation: {'enabled': False}
1|5|Loss: 0.90829998254776: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 5/5 [06:31<00:00, 77.67s/it]
INFO:torchtune.utils._logging:Saving checkpoint. This may take some time. Retrieving full model state dict...
INFO:torchtune.utils._logging:Getting full model state dict took 185.25 secs
INFO:torchtune.utils._logging:Model checkpoint of size 4.60 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0001_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0002_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 5.00 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0003_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.97 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0004_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0005_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0006_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0007_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 5.00 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0008_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.97 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0009_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0010_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0011_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0012_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 5.00 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0013_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.97 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0014_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0015_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0016_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0017_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 5.00 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0018_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.97 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0019_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0020_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0021_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0022_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 5.00 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0023_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.97 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0024_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0025_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0026_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0027_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 5.00 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0028_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.97 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0029_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0030_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0031_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0032_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 5.00 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0033_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.97 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0034_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0035_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0036_0.pt
INFO:torchtune.utils._logging:Model checkpoint of size 4.88 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0037_0.pt
INFO:torchtune.utils._logging:Adapter checkpoint of size 0.26 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/adapter_0.pt
WARNING:torchtune.utils._logging:Saving Llama3.2 Vision adapter weights to PEFT format is not supported, saving to torchtune format instead
WARNING:torchtune.utils._logging:PEFT integration for Llama3.2 Vision is not supported, skipping adapter config save
INFO:torchtune.utils._logging:Saving final epoch checkpoint.
INFO:torchtune.utils._logging:The full model checkpoint, including all weights and configurations, has been saved successfully.You can now use this checkpoint for further training or inference.
INFO:torchtune.utils._logging:Saving checkpoint took 450.86 secs
1|5|Loss: 0.90829998254776: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████

UX

If your function changed a public API, please add a dummy example of what the user experience will look like when calling it.
Here is a docstring example
and a tutorial example

I did not change any public API
I have added an example to docs or docstrings

pytorch-bot · 2024-11-12T20:13:11Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/torchtune/1994

📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit 3cdfd92 with merge base 4df97ad ():
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

ebsmothers · 2024-11-12T21:08:54Z

torchtune/models/llama3_2_vision/_model_builders.py

@@ -193,6 +193,9 @@ def lora_llama3_2_vision_11b(
        lora_dropout=lora_dropout,
        use_dora=use_dora,
        quantize_base=quantize_base,
+        # Update scaler block size to ensure that weights can be quantized evenly across 1, 2, 4, 6, 8 GPUs.
+        # This is dependent on ``clip_embed_dim`` so if that is updated, this variable should be as well
+        scaler_block_size=200 if quantize_base else None,


So no negative perf impact to using this value on < 8 GPUs?

With diff scaler block size

(joe-torchtune) [[email protected] ~/projects/joe-torchtune (update-clip-with-quant-nums)]$ tune run --nproc-per-node 4 lora_finetune_distributed --config llama3_2_vision/90B_qlora max_steps_per_epoch=5 lr_scheduler.num_warmup_steps=0 Running with torchrun... W1112 18:35:18.602000 4067602 site-packages/torch/distributed/run.py:793] W1112 18:35:18.602000 4067602 site-packages/torch/distributed/run.py:793] ***************************************** W1112 18:35:18.602000 4067602 site-packages/torch/distributed/run.py:793] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. W1112 18:35:18.602000 4067602 site-packages/torch/distributed/run.py:793] ***************************************** _component_: torchtune.modules.loss.CEWithChunkedOutputLoss lr_scheduler: _component_: torchtune.training.lr_schedulers.get_cosine_schedule_with_warmup num_warmup_steps: 0 max_steps_per_epoch: 5 metric_logger: _component_: torchtune.training.metric_logging.DiskLogger log_dir: /tmp/Llama-3.2-90B-Vision-Instruct/logs model: _component_: torchtune.models.llama3_2_vision.qlora_llama3_2_vision_90b apply_lora_to_mlp: true apply_lora_to_output: false decoder_trainable: frozen encoder_trainable: lora fusion_trainable: lora image_size: 560 lora_alpha: 16 lora_attn_modules: - q_proj - v_proj - output_proj lora_dropout: 0.0 lora_rank: 8 optimizer: _component_: torch.optim.AdamW fused: true lr: 0.0001 weight_decay: 0.01 output_dir: /tmp/qlora-llama3.2-vision-finetune profiler: _component_: torchtune.training.setup_torch_profiler active_steps: 2 cpu: true cuda: true enabled: false num_cycles: 1 output_dir: /tmp/qlora-llama3.2-vision-finetune/profiling_outputs profile_memory: false record_shapes: true wait_steps: 5 warmup_steps: 3 with_flops: false with_stack: false resume_from_checkpoint: false save_adapter_weights_only: false seed: null shuffle: true tokenizer: _component_: torchtune.models.llama3_2_vision.llama3_2_vision_transform image_size: 560 max_seq_len: 8192 path: /tmp/Llama-3.2-90B-Vision-Instruct/original/tokenizer.model NCCL version 2.21.5+cuda12.4 DEBUG:torchtune.utils._logging:Setting manual seed to local seed 3681342200. Local seed is seed + rank = 3681342200 + 0 Writing logs to /tmp/Llama-3.2-90B-Vision-Instruct/logs/log_1731465330.txt INFO:torchtune.utils._logging:FSDP is enabled. Instantiating model and loading checkpoint on Rank 0 ... INFO:torchtune.utils._logging:Instantiating model and loading checkpoint took 128.29 secs INFO:torchtune.utils._logging:Memory stats after model init: GPU peak memory allocation: 13.40 GiB GPU peak memory reserved: 14.53 GiB GPU peak memory active: 13.40 GiB INFO:torchtune.utils._logging:Optimizer is initialized. INFO:torchtune.utils._logging:Loss is initialized. INFO:torchtune.utils._logging:Dataset and Sampler are initialized. INFO:torchtune.utils._logging:Learning rate scheduler is initialized. WARNING:torchtune.utils._logging: Profiling disabled. INFO:torchtune.utils._logging: Profiler config after instantiation: {'enabled': False} 1|5|Loss: 0.8903459906578064: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 5/5 [13:04<00:00, 156.15s/it] INFO:torchtune.utils._logging:Saving checkpoint. This may take some time. Retrieving full model state dict... INFO:torchtune.utils._logging:Getting full model state dict took 207.90 secs INFO:torchtune.utils._logging:Model checkpoint of size 4.60 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0001_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0002_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 5.00 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0003_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 4.97 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0004_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0005_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0006_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0007_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 5.00 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0008_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 4.97 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0009_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0010_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0011_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0012_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 5.00 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0013_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 4.97 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0014_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0015_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0016_0.pt INFO:torchtune.utils._logging:Model checkpoint of size 4.66 GB saved to /tmp/Llama-3.2-90B-Vision-Instruct/hf_model_0017_0.pt

With default scaler

(joe-torchtune) [[email protected] ~/projects/joe-torchtune (update-clip-with-quant-nums)]$ tune run --nproc-per-node 4 lora_finetune_distributed --config llama3_2_vision/90B_qlora max_steps_per_epoch=5 lr_scheduler.num_warmup_steps=0 Running with torchrun... W1112 19:01:59.371000 237246 site-packages/torch/distributed/run.py:793] W1112 19:01:59.371000 237246 site-packages/torch/distributed/run.py:793] ***************************************** W1112 19:01:59.371000 237246 site-packages/torch/distributed/run.py:793] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. W1112 19:01:59.371000 237246 site-packages/torch/distributed/run.py:793] ***************************************** _component_: torchtune.modules.loss.CEWithChunkedOutputLoss lr_scheduler: _component_: torchtune.training.lr_schedulers.get_cosine_schedule_with_warmup num_warmup_steps: 0 max_steps_per_epoch: 5 metric_logger: _component_: torchtune.training.metric_logging.DiskLogger log_dir: /tmp/Llama-3.2-90B-Vision-Instruct/logs model: _component_: torchtune.models.llama3_2_vision.qlora_llama3_2_vision_90b apply_lora_to_mlp: true apply_lora_to_output: false decoder_trainable: frozen encoder_trainable: lora fusion_trainable: lora image_size: 560 lora_alpha: 16 lora_attn_modules: - q_proj - v_proj - output_proj lora_dropout: 0.0 lora_rank: 8 optimizer: _component_: torch.optim.AdamW fused: true lr: 0.0001 weight_decay: 0.01 output_dir: /tmp/qlora-llama3.2-vision-finetune profiler: _component_: torchtune.training.setup_torch_profiler active_steps: 2 cpu: true cuda: true enabled: false num_cycles: 1 output_dir: /tmp/qlora-llama3.2-vision-finetune/profiling_outputs profile_memory: false record_shapes: true wait_steps: 5 warmup_steps: 3 with_flops: false with_stack: false resume_from_checkpoint: false save_adapter_weights_only: false seed: null shuffle: true tokenizer: _component_: torchtune.models.llama3_2_vision.llama3_2_vision_transform image_size: 560 max_seq_len: 8192 path: /tmp/Llama-3.2-90B-Vision-Instruct/original/tokenizer.model NCCL version 2.21.5+cuda12.4 DEBUG:torchtune.utils._logging:Setting manual seed to local seed 1189370465. Local seed is seed + rank = 1189370465 + 0 Writing logs to /tmp/Llama-3.2-90B-Vision-Instruct/logs/log_1731466932.txt INFO:torchtune.utils._logging:FSDP is enabled. Instantiating model and loading checkpoint on Rank 0 ... INFO:torchtune.utils._logging:Instantiating model and loading checkpoint took 127.39 secs INFO:torchtune.utils._logging:Memory stats after model init: GPU peak memory allocation: 13.40 GiB GPU peak memory reserved: 14.53 GiB GPU peak memory active: 13.40 GiB INFO:torchtune.utils._logging:Optimizer is initialized. INFO:torchtune.utils._logging:Loss is initialized. INFO:torchtune.utils._logging:Dataset and Sampler are initialized. INFO:torchtune.utils._logging:Learning rate scheduler is initialized. WARNING:torchtune.utils._logging: Profiling disabled. INFO:torchtune.utils._logging: Profiler config after instantiation: {'enabled': False} 1|5|Loss: 0.8870511651039124: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 5/5 [13:37<00:00, 162.95s/it] INFO:torchtune.utils._logging:Saving checkpoint. This may take some time. Retrieving full model state dict... INFO:torchtune.utils._logging:Getting full model state dict took 201.79 secs

Looks to be no difference in speed or memory

Pass quantization_kwargs to CLIP builders

3cdfd92

facebook-github-bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Nov 12, 2024

joecummings requested review from pbontrager and ebsmothers November 12, 2024 20:50

ebsmothers reviewed Nov 12, 2024

View reviewed changes

ebsmothers approved these changes Nov 12, 2024

View reviewed changes

joecummings merged commit 51b31c8 into pytorch:main Nov 13, 2024
17 checks passed

joecummings deleted the update-clip-with-quant-nums branch November 13, 2024 03:26

joecummings added a commit that referenced this pull request Nov 13, 2024

Pass quantization_kwargs to CLIP builders (#1994)

19bd5d6

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Pass quantization_kwargs to CLIP builders #1994

Pass quantization_kwargs to CLIP builders #1994

Uh oh!

joecummings commented Nov 12, 2024 •

edited

Loading

Uh oh!

pytorch-bot bot commented Nov 12, 2024 •

edited

Loading

Uh oh!

ebsmothers Nov 12, 2024

Uh oh!

joecummings Nov 13, 2024 •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

Pass quantization_kwargs to CLIP builders #1994

Pass quantization_kwargs to CLIP builders #1994

Uh oh!

Conversation

joecummings commented Nov 12, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Context

Changelog

Test plan

UX

Uh oh!

pytorch-bot bot commented Nov 12, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/torchtune/1994

✅ No Failures

Uh oh!

ebsmothers Nov 12, 2024

Choose a reason for hiding this comment

Uh oh!

joecummings Nov 13, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

joecummings commented Nov 12, 2024 •

edited

Loading

pytorch-bot bot commented Nov 12, 2024 •

edited

Loading

joecummings Nov 13, 2024 •

edited

Loading