Commit History
-
Install fusion-embedding from PyPI
Verified signature120d15b -
Add Engram playground link
Verified signature5f3eeb5 -
Add Engram ecosystem section
Verified signaturef6d56b2
-
Card: separate model citation (Tonmoy) from paper citation (all authors)
Verified signaturec1bbeca -
Cite the arXiv paper (2607.18666)
Verified signatured94e3f9
-
inference.py: native video preprocessing (no qwen-vl-utils)
Verified signatureb551ea8 -
embed_video: native preprocessing (frame tensors and torchcodec path decode); drops the qwen-vl-utils dependency; bitwise-equal to the previous implementation and the raw base on identical inputs
Verified signatured632d6b -
inference.py: embed_video (base's own video path; purely additive)
Verified signaturee63b6c0 -
Add embed_video: the frozen base's own video path (qwen-vl-utils preprocessing, official defaults), gate-guarded like images; bitwise-verified against the base and the reference loader
Verified signaturef1a36c5
-
Card: positioning alt text matches the trained-parameters axis
Verified signatureb82ccac -
Positioning figure: trained-parameters axis (matches the technical report)
Verified signature415eb5c -
Remote code: batched embed_text_batch / embed_audio_batch entry points
Verified signaturee6d91bc -
AutoModel + trust_remote_code support (remote-code loading; existing loaders unchanged)
Verified signature27e1dd7 -
model card: positioning chart in cross-modal section
Verified signature7898f53 -
assets: unified-model positioning chart
Verified signaturece32ae4 -
model card: model overview figure
Verified signature72d9220 -
assets: model overview figure (brand palette)
Verified signaturecf18f7c
-
Note the safetensors/param-count meaning
Verified signaturefdb9184 -
Connector weights as safetensors (16.4M params)
Verified signature9592473 -
Add badges under the banner
Verified signatured38ccf1 -
Display banner image on the model card
Verified signatureef18a87 -
Add tagline banner to model card
Verified signature67e0464 -
MAEB: exact FSDKaggle2019 overlap numbers (608/4481 verified by Freesound id)
Verified signature87f90d8 -
MAEB: FSD2019Kaggle disclosure (anonymized filenames, disjointness unverifiable)
Verified signaturea65af61
-
Add banner image
Verified signature22f33a5 -
Add banner image
Verified signaturebcbb63c -
Add banner to model card
Verified signature2e9648e -
Add banner image
Verified signature6bd45fd -
Restructure card: architecture up top, unified-model comparison leads evaluation
Verified signature5020012 -
MAEB(beta) results summary
Verified signature15cd6b4 -
Add Gemini Embedding 2 baseline to cross-modal comparison
Verified signature49f04ad -
Add LanguageBind baseline to cross-modal comparison
Verified signatureef02ac0 -
v0.3 model card with three-version comparison
Verified signaturef263692 -
v0.3 config
Verified signature8e5ae97 -
v0.3: AudioCaps in-domain fine-tune stage (a2t R@10 0.741, R@1 0.332)
Verified signaturee92d9df
-
Model card: architecture diagram; fix bank size to 484K (v0.2)
Verified signaturebf1eb1a -
Add architecture diagram
Verified signatureabe5432
-
Document version tags
Verified signatureb82df83 -
from_pretrained: revision pinning
Verified signaturea815bcd -
v0.2 retrieval examples in model card
Verified signature37f999f -
v0.2 audio-to-image retrieval gallery
Verified signature11023b6 -
Bold best-per-column only in comparison tables
Verified signaturec35a99c -
Bold v0.2 numbers in comparison tables
Verified signaturec72602c -
v0.2 model card with v0.1 comparison
Verified signature889e950 -
v0.2 config
Verified signatureefa97da -
v0.2: 484K-pair corpus checkpoint (AudioCaps 0.717, Clotho-ZS 0.448)
Verified signature4ba4d39 -
Add VGGSound attribution to gallery
Verified signatureafc6633 -
Embed retrieval gallery in model card
Verified signature054c8a8 -
Add audio-to-image retrieval gallery
Verified signaturee085ecc -
Add full cross-modal metrics and qualitative retrieval examples
Verified signature5874cec