INDEX

Explanations

words related to extreme or intense behavior or characteristics

oai_token-act-pair · gpt-3.5-turbo

expressions of intensity, particularly those emphasizing extreme qualities or reactions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

tan

-0.75

itu

-0.73

illes

-0.72

raviolet

-0.70

activation

-0.68

ividual

-0.68

cel

-0.67

OTOS

-0.67

oral

-0.66

yer

-0.66

POSITIVE LOGITS

 wildly

1.01

ishly

0.85

efully

0.78

 inaccurate

0.73

 inco

0.73

 fluct

0.71

err

0.69

uously

0.69

 uncontroll

0.68

 flirt

0.67

Activations Density 0.007%

words related to extreme or intense behavior or characteristics

expressions of intensity, particularly those emphasizing extreme qualities or reactions

No Comments

No Known Activations