INDEX

Explanations

references to soap

oai_token-act-pair · gpt-3.5-turbo

references to soap operas

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

umbered

-0.73

etermined

-0.71

trl

-0.66

uding

-0.64

 omission

-0.63

alez

-0.63

orders

-0.62

 sworn

-0.60

CVE

-0.60

places

-0.60

POSITIVE LOGITS

 opera

1.14

 oper

1.09

 soap

0.96

eless

0.87

wich

0.84

bowl

0.81

pering

0.78

herer

0.78

utical

0.75

 Leone

0.75

Activations Density 0.009%

references to soap

references to soap operas

No Comments

No Known Activations