INDEX

Explanations

surnames or last names

oai_token-act-pair · gpt-3.5-turbo

proper nouns or significant names in a narrative context

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

BRE

-0.57

 rainbow

-0.57

 receptive

-0.56

 fetch

-0.56

uminati

-0.55

initions

-0.55

CFL

-0.54

 therapeutic

-0.53

 impair

-0.53

 Celest

-0.53

POSITIVE LOGITS

enegger

0.79

>.

0.76

 said

0.74

 told

0.74

 replied

0.74

(@

0.73

 joked

0.72

 says

0.71

 Says

0.71

 rhet

0.69

Activations Density 0.330%

surnames or last names

proper nouns or significant names in a narrative context

No Comments

No Known Activations