INDEX

Explanations

words related to negative or controversial actions or situations

oai_token-act-pair · gpt-3.5-turbo

terms related to reputation and its implications

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

sky

-0.65

nova

-0.65

Nig

-0.63

thur

-0.60

 veins

-0.60

danger

-0.60

gray

-0.60

erb

-0.59

Mig

-0.59

 Sherman

-0.59

POSITIVE LOGITS

ction

0.87

enment

0.80

ndum

0.79

ctions

0.78

ance

0.76

eval

0.76

issance

0.76

atives

0.75

ENCY

0.73

essed

0.73

Activations Density 0.164%

words related to negative or controversial actions or situations

terms related to reputation and its implications

No Comments

No Known Activations