INDEX

Explanations

references to awards or titles related to achievement

oai_token-act-pair · gpt-3.5-turbo

references to annual awards in sports or entertainment

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

pload

-0.65

express

-0.62

RANT

-0.59

 ambig

-0.57

grave

-0.55

 vulner

-0.55

amber

-0.53

ova

-0.53

 barr

-0.53

VAT

-0.53

POSITIVE LOGITS

ĳå£«

0.80

elight

0.72

 =================================

0.66

;;;;;;;;;;;;

0.59

itles

0.59

 Reporting

0.59

amines

0.57

 Assist

0.57

âĶĢâĶĢ

0.57

----

0.56

Activations Density 0.177%

references to awards or titles related to achievement

references to annual awards in sports or entertainment

No Comments

No Known Activations