INDEX

Explanations

terms related to news headlines or captions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Close" in the context of image captions or articles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ä»£

-0.72

ULE

-0.68

 Gibbs

-0.64

Dob

-0.61

 includ

-0.61

jury

-0.59

OTT

-0.58

 Lerner

-0.58

MT

-0.58

oled

-0.58

POSITIVE LOGITS

 Close

1.26

Close

1.11

enhagen

0.80

nery

0.78

quartered

0.77

 Shots

0.75

close

0.72

 Dragonbound

0.72

querque

0.72

 Shooting

0.71

Activations Density 0.007%

terms related to news headlines or captions

instances of the word "Close" in the context of image captions or articles

No Comments

No Known Activations