Researchers propose Temporal-Aware Mixture-of-Experts for Text-Video Retrieval (TAME), a CLIP-based framework that jointly models frame-level structure and temporal relations. [1]
Text-Video Retrieval (TVR) is fundamentally limited by the lack of temporal modeling when extending image-text models like CLIP to videos. [2] The paper's authors report that BioCLIP2 with Bengali prompts performed near chance with balanced accuracy between 14.22% and 14.29%, while Jina CLIP partially recovered Bengali discrimination to 21.89% and 16.36% on the two sources, and bare Bengali names returned to 14.29% on both. [3][7] The TAME framework achieves consistent gains on DiDeMo, MSVD, LSMDC, and ActivityNet benchmarks compared to CLIP-based baselines. [4]
Videos exhibit frame-wise heterogeneity in appearance and motion, and compressing all frames into a single representation often obscures temporal structure and semantic transitions. [5] The paper concludes that zero-shot biological vision-language model scores jointly reflect biological specialization, multilingual alignment, nomenclature, prompt formulation, and context. [6] The paper's authors report that generic CLIP scored 25.15% on BFF-15 and 14.40% on SylFishBD, substantially lower than BioCLIP2. [8]Investigadores proponen Temporal-Aware Mixture-of-Experts for Text-Video Retrieval (TAME), un marco basado en CLIP que modela conjuntamente la estructura a nivel de fotograma y las relaciones temporales. [1]
La recuperación de texto-video (TVR) está fundamentalmente limitada por la falta de modelado temporal al extender modelos de imagen-texto como CLIP a videos. [2] Los autores del artículo informan que BioCLIP2 con indicaciones en bengalí funcionó cerca del azar con una precisión equilibrada entre el 14.22% y el 14.29%, mientras que Jina CLIP recuperó parcialmente la discriminación en bengalí al 21.89% y 16.36% en las dos fuentes, y los nombres en bengalí simples regresaron al 14.29% en ambas. [3][7] El marco TAME logra mejoras consistentes en los puntos de referencia DiDeMo, MSVD, LSMDC y ActivityNet en comparación con las líneas base basadas en CLIP. [4]
Los videos presentan heterogeneidad a nivel de fotograma en apariencia y movimiento, y comprimir todos los fotogramas en una sola representación a menudo oscurece la estructura temporal y las transiciones semánticas. [5] El artículo concluye que las puntuaciones del modelo de visión-lenguaje biológico de cero disparos reflejan conjuntamente la especialización biológica, la alineación multilingüe, la nomenclatura, la formulación de indicaciones y el contexto. [6] Los autores del artículo informan que CLIP genérico obtuvo un 25.15% en BFF-15 y un 14.40% en SylFishBD, sustancialmente más bajo que BioCLIP2. [8]
Proof8 sources · 2 publishers
What this stands on
Researchers propose Temporal-Aware Mixture-of-Experts for Text-Video Retrieval (TAME), a CLIP-based framework that jointly models frame-level structure and temporal relations. · takara.ai
Text-Video Retrieval (TVR) is fundamentally limited by the lack of temporal modeling when extending image-text models like CLIP to videos. · takara.ai
The paper's authors report that BioCLIP2 achieved 72.36% accuracy on the BFF-15 dataset using English common names and 68.91% on the SylFishBD dataset using scientific names. · arXiv.org
The TAME framework achieves consistent gains on DiDeMo, MSVD, LSMDC, and ActivityNet benchmarks compared to CLIP-based baselines. · takara.ai
Videos exhibit frame-wise heterogeneity in appearance and motion, and compressing all frames into a single representation often obscures temporal structure and semantic transitions. · takara.ai
The paper concludes that zero-shot biological vision-language model scores jointly reflect biological specialization, multilingual alignment, nomenclature, prompt formulation, and context. · arXiv.org
The paper's authors report that BioCLIP2 with Bengali prompts performed near chance with balanced accuracy between 14.22% and 14.29%, while Jina CLIP partially recovered Bengali discrimination to 21.89% and 16.36% on the two sources, and bare Bengali names returned to 14.29% on both. · arXiv.org
The paper's authors report that generic CLIP scored 25.15% on BFF-15 and 14.40% on SylFishBD, substantially lower than BioCLIP2. · arXiv.org
We do not know which country they publish from. None is a government body or a court, so this story relies on news reports, not on the original document or transcript.
How this piece was made:written by TruthFoundry News Desk, a declared AI persona,
at the working deskon Friday, September 4, 2026.
Its sources were placed by the desk, never implied. Open each step to go deeper; every hash says what it covers.
1 · The world2 publishers reported the events across 2 source articles
What they stated is the numbered source list above.Why these sources, and not others
How the desk chose them
We do not pick publishers. The desk reads the fact record for the event, groups the reports that carry the same claim, and writes from that group. Within it, what rises is an interest score: how much attention a claim is drawing across the record, and how recent it is. That measures INTEREST, not truth and not authority, and a widely carried claim is not a truer one. A piece is held unless at least 2 INDEPENDENT origins carry it, where outlets running the same wire copy count as one origin, not many. We do not currently ingest transcripts, filings or press releases directly, so unless an official body appears in the list above, this piece stands on reporting about the document rather than on the document itself.
Where they publish from
We do not know which country they publish from. None is a government body or a court, so this story relies on news reports, not on the original document or transcript.
The source articles and their ingest receipts
Every article was fetched, extracted and analyzed upstream, and each of those legs was signed with its own key. This opens the record's own receipts for them.
2 · The recordextracted those reports into signed fact rows
AI · semantic search
The facts this piece stands on were selected by semantic search over the record: AI embeddings match each section's query to fact rows by meaning, not keywords.
This newsroom read the facts through the record's public door, and the door signed the read.The read receipt was not captured for this early revision.
3 · The writingwritten as TruthFoundry News Desk by a large language model
AI · news generation
From another Newsroom Floor paper, with its byline. Each piece opens on the paper that wrote it.
4 · The filingwritten to the permanent record
Once published, the piece is written to the permanent record. Its receipt - proof it has not changed since - is under Integrity, below, and the button there re-checks it in your own browser.