Marathi
PulseAugur coverage of Marathi — every cluster mentioning Marathi across labs, papers, and developer communities, ranked by signal.
5 day(s) with sentiment data
-
Developer builds 40M-parameter multilingual AI for delivery riders in 28 minutes
A developer has created a small, multilingual language model designed to assist delivery riders, particularly in areas with poor or no network connectivity. This 40-million-parameter model, trained in under 30 minutes o…
-
New method enhances Marathi cross-lingual summarization
Researchers have developed a new approach for cross-lingual abstractive summarization for the Marathi language. By combining data from the Marathi MahaSum corpus with the PMIndiaSum dataset, they fine-tuned the IndicTra…
-
Multilingual LLMs Fail Cross-Script Entity Matching, Study Finds
A new research paper published on arXiv details how multilingual language models struggle with entity matching when names are presented in different scripts. The study found that models like Qwen2.5-7B exhibit significa…
-
LLM confidence signals degrade on local knowledge, persist with hallucinations
Researchers have investigated the confidence signals of large language models (LLMs) across different temporal stages of problem-solving, distinguishing between pre-solve "Feeling-of-Knowing" (FOK) and post-solve "Judge…
-
Gharasathi launches as a privacy-first household AI
Gharasathi is a new household AI designed to manage personal data like finances and photos locally, ensuring privacy. It runs inference on a Mac Studio, with a focus on transparent data handling and evidence-based AI ex…
-
New dataset quantifies AI capability transfer across 15 languages
A new dataset called Multilingual GSM-Symbolic has been introduced to better understand how AI model capabilities transfer across different languages. This dataset, containing 30,000 item-matched question-answer pairs i…
-
New MMTClinic benchmark tests LLMs on multilingual clinical time-series data
Researchers have introduced MMTClinic, a new benchmark designed to evaluate large language models (LLMs) on clinical time-series data. This benchmark incorporates text, medical images, and physiological signals, featuri…
-
New benchmarks reveal LLM struggles with Indic languages and translation mechanics · 3 sources tracked
Researchers have developed VakyArth, a new benchmark designed to evaluate the pragmatic competence of large language models (LLMs) specifically within Indic languages like Hindi, Punjabi, Tamil, and Malayalam. Initial f…
-
Marathi TTS improved with Llama 3.1 8B LoRA and IndicF5
A developer has detailed their process for creating a more natural-sounding Text-to-Speech (TTS) system for the Marathi language. The project involved cleaning and processing messy, code-switched Marathi text to train a…
-
New benchmark reveals cross-lingual bias in LLMs
A new benchmark called INCLUDE has been developed to assess socio-cultural biases in Large Language Models (LLMs) across various Indian languages. Current safety alignment for LLMs is primarily English-focused, leading …
-
New research explores advanced tokenization for LLMs, improving efficiency and performance · 4 sources tracked
Researchers are developing new methods for tokenizing text in large language models to improve efficiency and performance. One approach, SuTRA, focuses on morphological structure for morphologically rich languages like …
-
LLM pipeline preserves document structure in Marathi-to-English translation
Researchers have developed a novel framework for translating government documents from Marathi to English, specifically addressing the challenge of preserving document structure and formatting. This system integrates la…
-
New Marathi POS Tagging Dataset and BERT Models Released
Researchers have introduced L3Cube-MahaPOS, a new dataset for Marathi Part-of-Speech (POS) tagging, addressing the scarcity of annotated resources for the language. The dataset contains over 32,000 manually annotated se…
-
New N-VSSM Model Outperforms Claude Opus 4.5 in Long-Form Narrative Consistency
Researchers have developed NarrativeWorldBench, a new benchmark designed to evaluate large language models (LLMs) on their ability to maintain narrative consistency in long-form audio dramas. Current frontier LLMs strug…
-
New dataset AgriGov boosts AI for Indian farmers
Researchers have developed AgriGov, a new multilingual dataset aimed at improving AI tools for Indian farmers. The dataset focuses on government schemes and welfare policies, initially covering 50 schemes across English…
-
New Marathi dataset BhashaSetu boosts low-resource translation quality
Researchers have introduced BhashaSetu, a new dataset designed to improve low-resource machine translation for Marathi. The dataset contains 2.78 million sentence pairs across various domains, including stemmed and lemm…
-
LLMs improve multilingual speech correction by tuning for fluency
Researchers have developed a new method for correcting disfluencies in multilingual speech transcripts using large language models (LLMs). The pipeline first identifies disfluent tokens and then uses these signals to fi…