David Faragó

dball's activity

• All
• Models
• Datasets
• Spaces
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes

commented 4 papers about 2 months ago

CantTalkAboutThis: Aligning Language Models to Stay on Topic in Dialogues

Paper • 2404.03820 • Published Apr 4 • 21 •

4

ReFT: Representation Finetuning for Language Models

Paper • 2404.03592 • Published Apr 4 • 74 •

17

Mixture-of-Depths: Dynamically allocating compute in transformer-based language models

Paper • 2404.02258 • Published Apr 2 • 102 •

5

Training LLMs over Neurally Compressed Text

Paper • 2404.03626 • Published Apr 4 • 21 •

2

commented 2 papers 3 months ago

Self-Rewarding Language Models

Paper • 2401.10020 • Published Jan 18 • 135 •

15

Self-Consuming Generative Models Go MAD

Paper • 2307.01850 • Published Jul 4, 2023 • 1 •

1

New activity in dball/zephyr-7b-sft-qlora 4 months ago

Is the drop in many metrics expected? Why do SFT first if it makes the model worse? Why not do DPO directly on the mistral model?

1

#1 opened 4 months ago by dball

New activity in dball/zephyr-7b-dpo-qlora 4 months ago

Is the high drop in GSM8K usual?

1

#1 opened 4 months ago by dball

New activity in teknium/CollectiveCognition-v1.1-Mistral-7B 5 months ago

Training code?

4

#5 opened 8 months ago by umarmnaq

New activity in malhajar/Mistral-7B-v0.2-meditron-turkish 5 months ago

Training script?

1

#1 opened 5 months ago by dball

New activity in fxmarty/tiny-llama-fast-tokenizer 10 months ago

Bugs when loading tokenizer or model

1

#3 opened 10 months ago by dball

New activity in Yhyu13/oasst-rlhf-2-llama-30b-7k-steps-hf 11 months ago

How did you merge?

1

#1 opened 11 months ago by dball

New activity in garage-bAInd/GPlatty-30B 11 months ago

Curious about model design

2

#2 opened 11 months ago by surya-narayanan