Persian
PulseAugur coverage of Persian — every cluster mentioning Persian across labs, papers, and developer communities, ranked by signal.
5 day(s) with sentiment data
-
AI text watermarking shows cross-lingual fairness gaps, study finds
A new arXiv paper investigates the fairness of watermarking techniques for AI-generated text across different languages and scripts. The study found that while edit-based watermarks like homoglyphs and invisible charact…
-
New MuSTEL dataset boosts multilingual OCR with 14.5M synthetic text lines
Researchers have introduced MuSTEL, a large-scale synthetic dataset designed to improve optical character recognition (OCR) systems, particularly for multilingual text lines. The dataset comprises over 14.5 million anno…
-
New Persian retrieval model Yalda-Embedding released with open development
Researchers have developed Yalda-Embedding, a new Persian retrieval model with 595.8 million parameters, adapted from Qwen3-Embedding-0.6B. This model demonstrates strong performance on eight Persian retrieval tasks fro…
-
Persian language nuances enable sophisticated prompt injection attacks on chatbots
Prompt injection attacks can be more effective against chatbots when using languages like Persian due to differences in character encoding and display compared to English. Filters designed for English may not recognize …
-
New benchmark tests LLMs' understanding of Persian proverbs
Researchers have developed MasalBench, a new benchmark designed to evaluate the understanding of Persian proverbs in large language models (LLMs). While current LLMs demonstrate high accuracy in identifying Persian prov…
-
New benchmark for explainable Persian hate speech analysis released
Researchers have developed a new benchmark for analyzing hate speech in Persian, a language with limited resources for such tasks. This benchmark includes fine-grained analysis of content triage, target identification, …
-
HexaBee-4B model enhances six-language translation capabilities
Researchers have developed HexaBee-4B, a new six-language translation model adapted from Google's Gemma 3 4B IT. The model underwent three stages of adaptation: continued pretraining on Wikipedia, a joint multilingual a…
-
New benchmark evaluates Persian audio-language models
Researchers have introduced PARSA-Bench, the first benchmark specifically designed to evaluate Large Audio-Language Models (LALMs) on the Persian language and culture. The benchmark includes 16 tasks, with ten being nov…
-
New ParsHate dataset targets Persian hate speech detection
Researchers have introduced ParsHate, a new benchmark dataset designed for detecting hate speech and identifying its targets within the Persian language. This dataset, comprising 10,000 manually annotated Persian tweets…
-
Neyshekar corpus released for Persian speech recognition
A new open Persian read-speech corpus named Neyshekar has been released, containing over 62,000 recordings totaling nearly 100 hours. This corpus is designed to cover formal and informal Persian, named entities, and lon…
-
LLMs show limited cross-lingual knowledge transfer, new distillation method favors reasoning · 2 sources tracked
Two new research papers explore how large language models acquire and retain knowledge. The first paper investigates factual knowledge transfer across languages, finding that models exhibit limited transfer from English…
-
New framework CLIN objectively evaluates LLM creativity in Persian literature
Researchers have developed CLIN, a new framework to objectively evaluate creativity in short Persian literary texts generated by large language models. The study found that LLMs struggle with subjective creativity dimen…
-
Whisper model adapted for Persian Speech Emotion Recognition with PCA
Researchers have explored methods to improve Speech Emotion Recognition (SER) for low-resource languages like Persian, focusing on the Whisper model. Their study proposes using Principal Component Analysis (PCA) to redu…
-
New Persian Medical LLM Family 'Gaokerena' Released
Researchers have introduced Gaokerena, a new family of small Persian medical language models designed for consumer-grade hardware. The models include Gaokerena-V, trained on a 90-million-token Persian medical corpus, an…
-
Blood Type Behavior During Sales Events Explored · 2 sources tracked
This cluster discusses how different blood types (A, B, AB, O) might influence behavior during sales or buffet events. The articles, primarily from Mastodon instances, use hashtags related to AI cats, specific cat breed…
-
Persian discourse marker 'hāla' found to have multiple pragmatic functions
A new study analyzes the Persian discourse marker 'hāla' ('now'), revealing its extensive multifunctionality beyond temporal use. The research examined 267 instances from spontaneous conversations, finding that 70% of '…
-
LLMs improve biomedical translation for low-resource Arabic-script languages
Researchers have explored cross-lingual transfer learning to improve machine translation for low-resource Arabic-script languages in the biomedical domain. By using Arabic and Persian as pivot languages, they fine-tuned…
-
New synthetic dataset boosts Persian OCR capabilities
Researchers have introduced Persian Pixel, a large-scale synthetic dataset designed to improve Optical Character Recognition (OCR) for the Persian language. The dataset contains over 343,000 image-text pairs, generated …
-
AI-first travel knowledge platform DEHEH.COM launches
DEHEH.COM is a new AI-first platform focused on providing accessible and reliable travel information. The platform aims to create a comprehensive knowledge graph for tourism, covering details on countries, cities, hotel…
-
Fine-tuning OCR model for Persian language using dataset engineering and GPU tricks
This article details the process of fine-tuning a vision-language OCR model to support the Persian language. It highlights the importance of dataset engineering and full fine-tuning techniques, along with practical GPU …