INDEX

Explanations

phrases or words related to identification, especially related to crime, suspects, and personal identification

oai_token-act-pair · gpt-3.5-turbo

references to identification and accountability 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

 overtake

-0.75

adder

-0.72

 brim

-0.69

 brainstorm

-0.68

 mull

-0.68

issance

-0.68

 rounds

-0.67

 paved

-0.67

 upkeep

-0.67

 prosper

-0.67

POSITIVE LOGITS

 Cosponsors

0.78

 è£ıè

0.75

 Individuals

0.74

imilar

0.73

 specific

0.71

ById

0.69

 perpetrators

0.69

vantage

0.68

Cause

0.66

Role

0.66

Activations Density 0.302%

phrases or words related to identification, especially related to crime, suspects, and personal identification

references to identification and accountability in various contexts

No Comments

No Known Activations