INDEX

Explanations

mentions of the word "arg."

oai_token-act-pair · gpt-3.5-turbo

occurrences of the term "arg" and its variations, suggesting a focus on arguments or debatable points

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 wash

-0.77

 seamless

-0.69

hua

-0.69

 uploaded

-0.68

 Fridays

-0.65

 crossover

-0.65

 overnight

-0.64

 maintenance

-0.63

 washed

-0.62

 trough

-0.62

POSITIVE LOGITS

arg

4.49

Arg

1.62

Arg

1.48

arg

1.44

argument

1.42

args

1.38

arge

1.30

arr

1.22

argon

1.20

ark

1.13

Activations Density 0.012%

mentions of the word "arg."

occurrences of the term "arg" and its variations, suggesting a focus on arguments or debatable points

No Comments

No Known Activations