added mlp and attn bias option to flash and paged llama models#85
Merged
added mlp and attn bias option to flash and paged llama models#85
Conversation
Signed-off-by: Joshua Rosenkranz <jmrosenk@us.ibm.com>
Signed-off-by: Joe Runde <Joseph.Runde@ibm.com>
njhill
approved these changes
May 6, 2024
Contributor
njhill
left a comment
There was a problem hiding this comment.
Thanks @JRosenkranz @joerunde!
openshift-merge-bot bot
referenced
this pull request
in red-hat-data-services/text-generation-inference
May 7, 2024
added mlp and attn bias option to flash and paged llama models (#85)
Xaenalt
pushed a commit
to Xaenalt/text-generation-inference
that referenced
this pull request
Aug 1, 2024
[pull] main from IBM:main
Xaenalt
pushed a commit
to Xaenalt/text-generation-inference
that referenced
this pull request
Aug 7, 2024
#### Motivation The `Calico` models currently set the mlp and attention bias to true, which was hard-coded to false in flash and paged llama implementations. This will use the config params set in huggingface/transformers#30031 to set those values properly. #### Modifications - added attention_bias, mlp_bias to config for Flash and Paged Llama implementations (default is False) - set bias in attention and mlp to the config value #### Result Models should be able to load properly if containing attention and mlp bias --------- Signed-off-by: Joshua Rosenkranz <jmrosenk@us.ibm.com> Signed-off-by: Joe Runde <Joseph.Runde@ibm.com> Co-authored-by: Joe Runde <Joseph.Runde@ibm.com>
Xaenalt
pushed a commit
to Xaenalt/text-generation-inference
that referenced
this pull request
Aug 12, 2024
#### Motivation The `Calico` models currently set the mlp and attention bias to true, which was hard-coded to false in flash and paged llama implementations. This will use the config params set in huggingface/transformers#30031 to set those values properly. #### Modifications - added attention_bias, mlp_bias to config for Flash and Paged Llama implementations (default is False) - set bias in attention and mlp to the config value #### Result Models should be able to load properly if containing attention and mlp bias --------- Signed-off-by: Joshua Rosenkranz <jmrosenk@us.ibm.com> Signed-off-by: Joe Runde <Joseph.Runde@ibm.com> Co-authored-by: Joe Runde <Joseph.Runde@ibm.com>
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
Motivation
[Describe why this change is needed]
The
Calicomodels currently set the mlp and attention bias to true, which was hard-coded to false in flash and paged llama implementations. This will use the config params set in huggingface/transformers#30031 to set those values properly.Modifications
[Describe the code changes]
Result
[Describe how the changes affects existing behavior and how to test it]
Models should be able to load properly if containing attention and mlp bias
Related Issues
NA