-
Notifications
You must be signed in to change notification settings - Fork 140
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
Showing
63 changed files
with
2,719 additions
and
317 deletions.
There are no files selected for viewing
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
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
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
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
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
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
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
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
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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,26 @@ | ||
| <!-- | ||
| Copyright (c) 2021 - present / Neuralmagic, Inc. All Rights Reserved. | ||
| Licensed under the Apache License, Version 2.0 (the "License"); | ||
| you may not use this file except in compliance with the License. | ||
| You may obtain a copy of the License at | ||
| http://www.apache.org/licenses/LICENSE-2.0 | ||
| Unless required by applicable law or agreed to in writing, | ||
| software distributed under the License is distributed on an "AS IS" BASIS, | ||
| WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | ||
| See the License for the specific language governing permissions and | ||
| limitations under the License. | ||
| --> | ||
|
|
||
| # CLIP Export Examples | ||
|
|
||
| The examples in `clip_onnx_export.py` provide the steps needed to export a CLIP model using sparseml's onnx exporting functionality. The models and pretrained weights are pulled in from [OpenClip](https://github.com/mlfoundations/open_clip/tree/main) and the command line tools provided allow exporting of a given model's Text and Visual branches. See the OpenClip repository for a full list of available models. For the CoCa models available in OpenClip, an additional text-decoder is also exported. | ||
|
|
||
| ## Installation | ||
|
|
||
| The examples provided require torch nighly and `open_clip_torch==2.20.0` to be installed. To work within the `sparseml` environment, be sure to set the environment variable `MAX_TORCH` to your installed version when | ||
| installing torch nightly. | ||
|
|
||
| Example: `MAX_TORCH="2.1.0.dev20230613+cpu"` |
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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,61 @@ | ||
| # Copyright (c) 2021 - present / Neuralmagic, Inc. All Rights Reserved. | ||
| # | ||
| # Licensed under the Apache License, Version 2.0 (the "License"); | ||
| # you may not use this file except in compliance with the License. | ||
| # You may obtain a copy of the License at | ||
| # | ||
| # http://www.apache.org/licenses/LICENSE-2.0 | ||
| # | ||
| # Unless required by applicable law or agreed to in writing, | ||
| # software distributed under the License is distributed on an "AS IS" BASIS, | ||
| # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | ||
| # See the License for the specific language governing permissions and | ||
| # limitations under the License. | ||
|
|
||
| import torch | ||
| import torch.nn as nn | ||
|
|
||
|
|
||
| class VisualModel(nn.Module): | ||
| def __init__(self, visual_model: torch.nn.Module, output_tokens: bool): | ||
|
|
||
| super().__init__() | ||
|
|
||
| self.visual_model = visual_model | ||
| self.visual_model.output_tokens = output_tokens | ||
|
|
||
| def forward(self, x): | ||
| return self.visual_model(x) | ||
|
|
||
|
|
||
| class TextModel(nn.Module): | ||
| def __init__( | ||
| self, | ||
| token_embedding: torch.nn.Embedding, | ||
| positional_embedding: torch.nn.parameter.Parameter, | ||
| transformer: torch.nn.Module, | ||
| ln_final: torch.nn.LayerNorm, | ||
| text_projection: torch.nn.parameter.Parameter, | ||
| attn_mask: torch.Tensor, | ||
| ): | ||
|
|
||
| super().__init__() | ||
|
|
||
| self.token_embedding = token_embedding | ||
| self.positional_embedding = positional_embedding | ||
| self.transformer = transformer | ||
| self.ln_final = ln_final | ||
| self.text_projection = text_projection | ||
| self.attn_mask = attn_mask | ||
| self.cast_dtype = self.transformer.get_cast_dtype() | ||
|
|
||
| def forward(self, input_ids): | ||
| x = self.token_embedding(input_ids).to(self.cast_dtype) | ||
| x = x + self.positional_embedding.to(self.cast_dtype) | ||
| x = x.permute(1, 0, 2) # NLD -> LND | ||
| x = self.transformer(x, attn_mask=self.attn_mask) | ||
| x = x.permute(1, 0, 2) # LND -> NLD | ||
| x = self.ln_final(x) # [batch_size, n_ctx, transformer.width] | ||
| # take features from the eot embedding (eot_token = highest in each sequence) | ||
| x = x[torch.arange(x.shape[0]), input_ids.argmax(dim=-1)] @ self.text_projection | ||
| return x |
Oops, something went wrong.