INDEX

Explanations

names, likely related to law enforcement or legal matters

oai_token-act-pair · gpt-3.5-turbo

proper nouns, specifically names of individuals

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

mble

-0.83

intendo

-0.70

 âĢº

-0.70

ymm

-0.68

ģĸ

-0.67

ãģ¯

-0.66

NES

-0.64

{}

-0.63

atomic

-0.62

ampunk

-0.61

POSITIVE LOGITS

kson

0.80

 himself

0.70

's

0.64

SON

0.62

zon

0.60

 Quin

0.58

 herself

0.57

 constituents

0.57

NG

0.55

 unworthy

0.55

Activations Density 0.370%

names, likely related to law enforcement or legal matters

proper nouns, specifically names of individuals

No Comments

No Known Activations