Search results

Add RAG Fine-tuning

Status: Open.

#35 In ai-hero/llm-research-fine-tuning;

· rparundekar opened on Jan 4, 2024

DPO

Status: Open.

#34 In ai-hero/llm-research-fine-tuning;

· rparundekar opened on Jan 4, 2024

How does the loss / cost compare with other frameworks for the same dataset

Status: Open.

#33 In ai-hero/llm-research-fine-tuning;

· rparundekar opened on Jan 4, 2024

LoRA is taking longer than SFT to train on A100s

Status: Open.

#32 In ai-hero/llm-research-fine-tuning;

· rparundekar opened on Jan 4, 2024

Fix Mistral 7b issue

Status: Open.

#31 In ai-hero/llm-research-fine-tuning;

· rparundekar opened on Jan 4, 2024

Create service using knative that scales to 0. Currently, the costs are crazy

Status: Open.

#30 In ai-hero/llm-research-fine-tuning;

· rparundekar opened on Dec 29, 2023

Add tags for dataset name, version, config, and training params to W&B

Status: Open.

#27 In ai-hero/llm-research-fine-tuning;

· rparundekar opened on Dec 25, 2023

Add eval logic based on https://txt.cohere.com/evaluating-llm-outputs/#llm-generated-evaluation

Status: Open.

#26 In ai-hero/llm-research-fine-tuning;

· rparundekar opened on Dec 25, 2023

Investigate if max seq length affects quality?

Status: Open.

#25 In ai-hero/llm-research-fine-tuning;

· rparundekar opened on Dec 25, 2023

Make test split eval work

Status: Open.

#21 In ai-hero/llm-research-fine-tuning;

· rparundekar opened on Dec 24, 2023

Save the model with the best eval score.

Status: Open.

#20 In ai-hero/llm-research-fine-tuning;

· rparundekar opened on Dec 24, 2023

Load base model from S3

Status: Open.

#18 In ai-hero/llm-research-fine-tuning;

· rparundekar opened on Dec 24, 2023