INDEX

Explanations

contractions indicating negation or questioning in written text

oai_token-act-pair · gpt-3.5-turbo

negative contractions in questions and statements

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Reviewer

-0.78

cele

-0.69

æ©

-0.63

itone

-0.63

Ñģ

-0.63

places

-0.59

 Cosponsors

-0.59

forms

-0.59

Tracker

-0.58

inen

-0.58

POSITIVE LOGITS

 bother

0.67

berra

0.66

ync

0.63

 gonna

0.62

ffff

0.62

igslist

0.61

estine

0.60

icable

0.60

 necessarily

0.60

attering

0.59

Activations Density 0.071%

contractions indicating negation or questioning in written text

negative contractions in questions and statements

No Comments

No Known Activations