INDEX

Explanations

the word "Bartlett"

oai_token-act-pair · gpt-3.5-turbo

mentions of the name "Bart."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

WARD

-0.64

ixties

-0.63

lled

-0.62

CFR

-0.61

 Gazette

-0.60

VEN

-0.59

 conception

-0.59

STD

-0.58

CLUS

-0.58

intage

-0.58

POSITIVE LOGITS

lett

0.99

ric

0.98

rys

0.98

oken

0.94

©¶æ¥µ

0.94

esian

0.93

 Bart

0.91

ram

0.89

olini

0.88

ritis

0.81

Activations Density 0.007%

the word "Bartlett"

mentions of the name "Bart."

No Comments

No Known Activations