INDEX

Explanations

phrases related to bravery or audacity

oai_token-act-pair · gpt-3.5-turbo

instances of the word “gall” and related forms, often in contexts suggesting boldness or audacity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

anyahu

-0.78

ĨĴ

-0.77

ioch

-0.74

 equity

-0.72

nil

-0.72

avez

-0.69

 indo

-0.67

 Deal

-0.66

 confir

-0.65

ensor

-0.65

POSITIVE LOGITS

 gall

1.05

 bladder

1.01

antly

0.89

assing

0.78

stones

0.78

ingly

0.77

ading

0.71

bl

0.71

fruit

0.71

draw

0.70

Activations Density 0.019%

phrases related to bravery or audacity

instances of the word “gall” and related forms, often in contexts suggesting boldness or audacity

No Comments

No Known Activations