INDEX

Explanations

names containing the specific pattern "Sar" followed by different characters

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of the name "Sar" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 post

-0.62

 grid

-0.62

MAP

-0.62

 tense

-0.62

 flourish

-0.60

ense

-0.59

 favorable

-0.59

 Hardy

-0.58

set

-0.58

 curfew

-0.58

POSITIVE LOGITS

kees

1.31

wark

1.10

andon

1.08

vati

1.07

aya

1.07

abeth

1.06

aband

1.04

uba

1.02

ashtra

0.97

ban

0.96

Activations Density 0.028%

names containing the specific pattern "Sar" followed by different characters

repeated mentions of the name "Sar" in various contexts

No Comments

No Known Activations