INDEX

Explanations

words related to strong emphasis or extreme situations

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "utter" and its variations, signaling strong expressions of emphasis or negativity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

amsung

-0.75

xual

-0.74

OHN

-0.71

 Feder

-0.69

actionDate

-0.69

oult

-0.68

llah

-0.66

morrow

-0.66

ioch

-0.66

pei

-0.65

POSITIVE LOGITS

ances

1.09

ance

0.96

 utter

0.86

most

0.85

ancing

0.75

ing

0.74

aneous

0.74

TY

0.73

ables

0.72

 amaz

0.71

Activations Density 0.016%

words related to strong emphasis or extreme situations

occurrences of the word "utter" and its variations, signaling strong expressions of emphasis or negativity

No Comments

No Known Activations