chk2 21 Jun

data/projects.mdx

@@ -16,10 +16,10 @@ title: Projects
 ## Recent
 
 <div className="grid sm:grid-cols-2 gap-6">
-	<ProjectWithBadges url="https://github.com/RomanKoshkin/transllama" title="🔪toLLMatch" badges={["NLP", "pytorch", "machine translation", "AWS"]}>
+	<ProjectWithBadges url="https://github.com/RomanKoshkin/toLLMatch" title="🔪toLLMatch" badges={["NLP", "pytorch", "machine translation", "AWS"]}>
   		Zero-shot context-aware simultaneous machine translation leveraging an open-source ASR engine and LLM.
 	</ProjectWithBadges>
-	<ProjectWithBadges url="https://github.com/RomanKoshkin/toLLMatch" title="🦙TransLLaMa" badges={["NLP", "LLMOps", "machine translation", "MLOps"]}>
+	<ProjectWithBadges url="https://github.com/RomanKoshkin/transllama" title="🦙TransLLaMa" badges={["NLP", "LLMOps", "machine translation", "MLOps"]}>
   		LLM-based simultaneous speech-to-text machine translation. Decoder-only large language models (LLMs) have recently demonstrated impressive capabilities in text generation and reasoning. Nonetheless, they have limited applications in simultaneous machine translation (SiMT), currently dominated by encoder-decoder transformers. This study demonstrates that, after fine-tuning on a small dataset comprising causally aligned source and target sentence pairs, a pre-trained open-source LLM can control input segmentation directly by generating a special "wait" token. This obviates the need for a separate policy and enables the LLM to perform English-German and English-Russian SiMT tasks with BLEU scores that are comparable to those of specific state-of-the-art baselines. We also evaluated closed-source models such as GPT-4, which displayed encouraging results in performing the SiMT task without prior training (zero-shot), indicating a promising avenue for enhancing future SiMT systems.
 	</ProjectWithBadges>
 	<ProjectWithBadges url="https://github.com/RomanKoshkin/conv-seq" title="🧠convSeq" badges={["DL", "computational neuroscience"]}>