INDEX

Explanations

phrases indicating a level of importance or relevance towards specific topics or issues

oai_token-act-pair · gpt-3.5-turbo

instances of the word "concerned"

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

artifacts

-0.85

Bom

-0.72

obs

-0.70

robe

-0.68

fruit

-0.68

arb

-0.67

buff

-0.67

ingen

-0.67

ingers

-0.66

guided

-0.65

POSITIVE LOGITS

 proble

0.73

 citiz

0.72

 trolling

0.70

atives

0.67

 Schr

0.67

reon

0.67

 Concern

0.67

cerned

0.66

NESS

0.66

ately

0.65

Activations Density 0.023%

phrases indicating a level of importance or relevance towards specific topics or issues

instances of the word "concerned"

No Comments

No Known Activations