INDEX

Explanations

phrases that are being rated

oai_token-act-pair · gpt-3.5-turbo

ratings and instances of being seen or witnessed

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Adin

-0.72

 Klein

-0.68

 Piper

-0.67

poke

-0.67

Vox

-0.64

 Nost

-0.62

 confess

-0.62

stick

-0.61

 Juliet

-0.60

 fart

-0.60

POSITIVE LOGITS

Rated

2.19

seen

1.91

gency

1.84

Fil

1.54

done

1.19

Film

1.16

rified

1.13

Dim

1.10

rification

1.08

rimp

1.01

Activations Density 0.058%

phrases that are being rated

ratings and instances of being seen or witnessed

No Comments

No Known Activations