Commit History
-
Card: note the integration files are on the kw-preview branch too
Verified signatureb91c1c2 -
Model card: add Sentence Transformers usage section
Verified signaturef496b0e -
Add Sentence Transformers integration (custom module)
Verified signaturef18df08
-
Evaluation integrity: report Common Voice under the training_datasets disclosure; remove internal notes
Verified signature811ad73 -
Add the evaluation-integrity section (Common Voice overlap disclosure)
Verified signature3422c99
-
Add self-hosted serving section (vLLM and SGLang plugins)
Verified signatureb25c552 -
Cite the corpus paper (arXiv:2608.01560) at the emotion-regression note it explains
Verified signaturece0a206
-
Add Tactus (touch sense pack) to the model family listing
Verified signature71b9676
-
Install fusion-embedding from PyPI
Verified signature8948f6a -
Add Engram playground link
Verified signature3bc1763 -
Add Engram ecosystem section
Verified signature9f242cb
-
Card: clean rebuild - single-row badges + RunPod on next row, fixed frontmatter
Verified signature66f6df9 -
Card: badges on one row, Deploy on RunPod on next row (centered)
Verified signature48a0ddb -
Card: fix line endings (clean LF; restore YAML frontmatter parsing)
Verified signature3f05159 -
Card: revert to dynamic RunPod badge
Verified signaturef998e1d -
Card: static Deploy on RunPod badge (always renders)
Verified signature695c3ab -
Card: restore banner, add RunPod deploy
Verified signature3cff2e1 -
Card: add RunPod deploy section
Verified signature02c954d -
Card: add RunPod Hub deploy badge + endpoint call instructions
Verified signature3ab0c1c
-
Add v0.3 banner to model card
Verified signature231f00e
-
v0.3-preview: retrieval-tuned flagship
Verified signatureAudioCaps a2t R@10 0.743 -> 0.785 over v0.2; t2a R@10 0.782. MAEB(audio-only) 19-task mean 0.454; leads IEMOCAP speaker-gender (0.938). RAVDESS zero-shot emotion regressed to 0.21. Non-audio (text/image/video) outputs remain bit-for-bit the frozen base model's, verified on this artifact before push. Ships a second keyword-tuned revision on branch v0.3-kw-preview.1720d8b
-
Card: separate model citation (Tonmoy) from paper citation (all authors)
Verified signature70d7728 -
Cite the arXiv paper (2607.18666)
Verified signature1903dc0
-
card: link the Ember thermal sense pack
Verified signature529b017 -
Card: MAEB nine-task section with the v0.2 rescore
Verified signaturev0.2 evaluated at revision 9451b840f0d1 (mteb 2.18.x, identical protocol to the v0.1 submission): mean .361 vs .354, six of nine tasks improved.eeb829b -
config.json: restore the remote-code loading schema
Verified signatureThe v0.2 weights commit replaced config.json with the packager payload, dropping auto_map/architectures and the model-construction fields the trust_remote_code path requires. Restored from the previous revision with the v0.2 benchmark payload merged in.9451b84 -
v0.2: fine-tune on the expanded AudioCaps 2.0 pool (86,394 pairs, 1,600 steps)
Verified signatureAudioCaps A2T R@10 0.743 -> 0.759, T2A R@10 0.775 -> 0.785; Clotho and VGGSound audio-text within noise of v0.1 under paired protocols; emergent audio-image up ~1.5 points. Non-audio outputs remain bit-for-bit the base model's (verified on this artifact before push). Card updated accordingly.35d1d2f
-
inference.py: native video preprocessing (no qwen-vl-utils)
Verified signature2d30494 -
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 signature1ee57f7 -
inference.py: embed_video (base's own video path, gate-guarded)
Verified signaturea177d26 -
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 signature85b38b4
-
Card: AudioCaps T2A R@1 predecessor cell now recorded (0.280)
Verified signature5154536 -
Card: positioning alt text matches the trained-parameters axis
Verified signaturec1af335 -
Positioning figure: trained-parameters axis (matches the technical report)
Verified signaturebe7764a -
inference.py: install URL points at the renamed family repo
Verified signatureb37f54b -
Card: code links to the renamed family repo (Eximius-Labs/fusion-embedding); Clotho T2A R@1 predecessor cell now recorded (0.136)
Verified signature2b5bcc8 -
AutoModel + trust_remote_code support (remote-code loading; existing loaders unchanged)
Verified signature26319a8 -
Card: precise cell count (8 of 12 release-protocol cells; every recorded text-to-audio direction)
Verified signaturef482f57 -
inference.py: gate guard on embed_image (non-audio inputs must run with the gate closed)
Verified signature4bad8f3 -
model card: positioning chart visible in Evaluation
Verified signature71170e9 -
assets: positioning chart — distinct hues for the two generations
Verified signature5f0502b -
model card: positioning chart in cross-modal section
Verified signature91572d8 -
assets: unified-model positioning chart
Verified signaturef4df188 -
assets: overview figure from the unified family generator
Verified signaturea3f1db1 -
assets: overview figure in Eximius Labs palette
Verified signaturea4c0ace -
assets: overview figure — EOS chip inside pooling box
Verified signature53c614a -
assets: overview figure — horizontal pooling arrow
Verified signature3550fd6 -
assets: model overview figure v2 (layer-stack style)
Verified signature267b7fe -
model card: add model overview figure
Verified signature59cd63a -
assets: model overview figure
Verified signature9fdb6b6