INDEX

Explanations

names of individuals accompanied by their professional titles

oai_token-act-pair · gpt-3.5-turbo

titles and roles of individuals often associated with authority or governance

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

etheless

-0.78

âĶģ

-0.74

âĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢ

-0.72

 âī¡

-0.71

âķĲ

-0.70

stream

-0.67

opp

-0.67

 grouped

-0.66

 fulfil

-0.65

 reap

-0.64

POSITIVE LOGITS

 Randy

1.40

 Jerry

1.37

 Kathy

1.37

 Larry

1.35

 Mike

1.34

 Darrell

1.34

 Steve

1.34

 Shawn

1.33

 Gary

1.33

 Cheryl

1.31

Activations Density 0.200%

names of individuals accompanied by their professional titles

titles and roles of individuals often associated with authority or governance

No Comments

No Known Activations