Add FCD scoring support and longform inference mode#1252
Open
vmendelev wants to merge 1 commit intopr3/magpie-tts-backendfrom
Open
Add FCD scoring support and longform inference mode#1252vmendelev wants to merge 1 commit intopr3/magpie-tts-backendfrom
vmendelev wants to merge 1 commit intopr3/magpie-tts-backendfrom
Conversation
- Add save_codes config option to MagpieTTS backend for codec code saving
- Encode codec .pt files as base64 in debug_info for transfer to client
- Add _save_codec_data() to VLLMMultimodalModel to decode and save codec
files on the client side (for FCD scoring pipeline)
- Update InferenceConfig to use ModelInferenceParameters (NeMo API change)
- Fix run_inference_on_dataset unpacking for updated NeMo API
- Add longform_mode config ("auto"/"always"/"never") for longform inference
- Add --save_codes and --longform_mode CLI arguments to serve_unified.py
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
vmendelev
force-pushed
the
pr6/fcd-scoring-longform
branch
from
087e8b8 to
20287bb
Compare
vmendelev
force-pushed
the
pr3/magpie-tts-backend
branch
from
f4bc5cd to
1a9f4fa
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
--longform_mode auto/always/never) for MagpieTTSModelInferenceParametersAPI forInferenceConfigrun_inference_on_datasetunpacking for updated NeMo APIHow FCD scoring works
magpie_tts_backend.py): Whensave_codes=True, saves predicted codec.ptfiles, encodes as base64 indebug_info["codec_data"]vllm_multimodal.py): Extractscodec_datafrom debug_info, decodes from base64, saves as.ptfile on disk, stores path indebug_info["codec_codes_path"]score.pyin PR Add nv_tts dataset and evaluation scripts #1248): Readscodec_codes_pathfrom output, symlinks codec files for FCD evaluationDepends on
Files changed
recipes/multimodal/server/backends/magpie_tts_backend.py— save_codes, longform_mode, ModelInferenceParameters APInemo_skills/inference/server/serve_unified.py—--save_codes,--longform_modeCLI argsnemo_skills/inference/model/vllm_multimodal.py— codec data saving on client sideTest plan
--save_codesproduces codec.ptfiles--longform_mode alwaysenables longform inference🤖 Generated with Claude Code