INDEX

Explanations

phrases indicating responsibility or assignment of blame

oai_token-act-pair · gpt-3.5-turbo

the word "to" and its variations indicating purpose or intention

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ById

-0.66

Dur

-0.65

 pops

-0.64

 deductions

-0.61

 didnt

-0.60

 scratches

-0.60

 reperto

-0.60

 Advertisement

-0.60

 fluctuations

-0.59

 scares

-0.58

POSITIVE LOGITS

asted

1.37

asting

1.28

asters

1.08

ying

1.04

wered

1.04

 blame

1.03

pless

1.00

othy

0.99

iling

0.97

ilet

0.97

Activations Density 0.127%

phrases indicating responsibility or assignment of blame

the word "to" and its variations indicating purpose or intention

No Comments

No Known Activations