INDEX

Explanations

words related to connections, associations, or relationships between different entities

oai_token-act-pair · gpt-3.5-turbo

references to connections or links in various contexts

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

 faced

-0.72

iae

-0.71

ŃĶ

-0.69

icer

-0.69

ctory

-0.67

ĨĴ

-0.67

Written

-0.67

gerald

-0.65

Ĥª

-0.64

į

-0.63

POSITIVE LOGITS

 facilitate

0.73

 shore

0.72

 ensure

0.70

 infinity

0.68

 haunt

0.67

 schedule

0.67

agate

0.67

 acquire

0.67

 conserve

0.64

 confirm

0.64

Activations Density 0.125%

words related to connections, associations, or relationships between different entities

references to connections or links in various contexts

No Comments

No Known Activations