Add LFM2.5 1.2B support with CI integration#17509
Open
larryliu0820 wants to merge 3 commits intomainfrom
Open
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/17509
Note: Links to docs will display an error until the docs builds have been completed. ❌ 5 New FailuresAs of commit f998ad1 with merge base b69cb70 ( NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
bot
added
the
CLA Signed
mergennachin
approved these changes
Feb 18, 2026
This adds support for exporting LiquidAI's LFM2.5-1.2B-Instruct model, which features improved training (28T tokens vs 10T) and extended context support (32K tokens). Changes: - Add lfm2_5_1_2b model configuration (examples/models/lfm2/config/lfm2_5_1_2b_config.json) - Register lfm2_5_1_2b in EXECUTORCH_DEFINED_MODELS and HUGGING_FACE_REPO_IDS - Add lfm2_5_1_2b to ModelType enum - Update README with LFM2.5 export examples and extended context usage - Add CI test for lfm2_5_1_2b in test_torchao_huggingface_checkpoints.sh - Add lfm2_5_1_2b to trunk.yml test matrix (XNNPACK and TorchAO backends) The model uses the same hybrid architecture as LFM2 (convolution + attention layers) with 16 layers and 8192 FFN dimension. Can be exported with custom max_seq_length and max_context_length for extended context support. Fixes #17439 Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
The CI was failing because the huggingface-cli command wasn't available. Changed to use 'hf download' to match qwen3_4b and phi_4_mini. Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
The convert_weights script couldn't be executed as a module because it was missing the main() function and __main__ block. Added argparse CLI interface matching other model converters (qwen3, phi_4_mini). Tested successfully: - Converted LFM2.5-1.2B-Instruct checkpoint (2.2GB output) - CLI arguments work correctly Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
larryliu0820
force-pushed
the
add-lfm2.5-support
branch
from
431a4df to
f998ad1
Compare
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
2 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.
This adds support for exporting LiquidAI's LFM2.5-1.2B-Instruct model, which features improved training (28T tokens vs 10T) and extended context support (32K tokens).
Changes:
The model uses the same hybrid architecture as LFM2 (convolution + attention layers) with 16 layers and 8192 FFN dimension. Can be exported with custom max_seq_length and max_context_length for extended context support.
Fixes #17439