INDEX

Explanations

phrases related to searching for information

oai_token-act-pair · gpt-3.5-turbo

phrases indicating inquiries or questions about trustworthiness in news and information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Lago

-0.82

è£

-0.71

anas

-0.71

verend

-0.71

umbn

-0.70

ocol

-0.67

fer

-0.66

SPONSORED

-0.64

arser

-0.62

Mub

-0.62

POSITIVE LOGITS

Looking

1.00

 suspic

0.88

 adolesc

0.84

 citiz

0.78

uez

0.78

allery

0.75

 Looking

0.71

 juven

0.71

 warr

0.70

 nodd

0.69

Activations Density 0.008%

phrases related to searching for information

phrases indicating inquiries or questions about trustworthiness in news and information

No Comments

No Known Activations