INDEX

Explanations

words related to specific names like "Hank" and "Clint"

oai_token-act-pair · gpt-3.5-turbo

names of individuals involved in various contexts or narratives

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

theless

-0.83

 à¤

-0.83

vic

-0.77

heed

-0.73

à¤

-0.72

lihood

-0.72

 describ

-0.64

mil

-0.64

ple

-0.64

cules

-0.63

POSITIVE LOGITS

 Barton

0.97

 Hank

0.83

alach

0.83

 Caldwell

0.77

 Crawford

0.76

 Hawks

0.74

icago

0.74

 Stark

0.72

 Pastor

0.70

ATA

0.67

Activations Density 0.015%

words related to specific names like "Hank" and "Clint"

names of individuals involved in various contexts or narratives

No Comments

No Known Activations