INDEX

Explanations

words related to credibility or plausibility

oai_token-act-pair · gpt-3.5-turbo

terms related to various forms of "ability" or "capability."

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

 wart

-0.59

 diver

-0.59

ART

-0.57

 disappoint

-0.57

nat

-0.57

mil

-0.56

ang

-0.56

cum

-0.55

 World

-0.54

tug

-0.54

POSITIVE LOGITS

ibility

4.85

ible

2.99

ibilities

2.86

ibly

2.71

ibles

2.64

IBLE

2.26

ibl

1.86

ability

1.80

BILITY

1.62

abilities

1.60

Activations Density 0.012%

words related to credibility or plausibility

terms related to various forms of "ability" or "capability."

No Comments

No Known Activations