INDEX

Explanations

phrases that indicate a place or a grouping of people

oai_token-act-pair · gpt-3.5-turbo

phrases that emphasize membership or prevalence in a category

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

culosis

-0.67

 accordingly

-0.66

 gaze

-0.60

 perpend

-0.60

hs

-0.59

 dispose

-0.59

 comply

-0.58

tion

-0.57

 mustache

-0.57

 understands

-0.55

POSITIVE LOGITS

inea

0.74

icial

0.72

elong

0.71

sted

0.67

onel

0.67

apest

0.65

odox

0.64

ãĤ¨

0.64

itol

0.64

icer

0.64

Activations Density 0.172%

phrases that indicate a place or a grouping of people

phrases that emphasize membership or prevalence in a category

No Comments

No Known Activations