Niranjan Akella

icon indicating a website link https://www.niranjanakella.com

icon location Hyderabad, India icon location Expertise in developing Large Language Models (LLM), fine-tuning them with PEFT-LoRA & also developing other predictive model for NLP & Computer Vision

Results 15 comments of Niranjan Akella

Support for T5 Architecture

@EricLBuehler Non-Quantized f16,32 implementation currently holds more precedence. But if possible, would also like to have a quantized implementation too. Also I wish to know if LoRA adapters can be...

Support for T5 Architecture

@EricLBuehler No problem sounds good. I am looking forward to trying it out soon.

Format conversion issue after downstream SFT

Hey the tokenizer.model is still missing from the official repo, unable to convert tl2 without it. @BradZhone

Amadeus Python SDK discrepancies in airport suggestions

@minjikarin @cbetta @anthonyroux Please let me know how I can solve this or if I am making some mistake in the way I am using this API.

Unable to process audio-video modality properly

@freddyaboulton Any idea how to host a local or bedrock LLM for this purpose like I am trying to do? Should have have a different session created with the audio...