INDEX

Explanations

abbreviated forms of words or phrases signifying emphasis or importance

oai_token-act-pair · gpt-3.5-turbo

references to online advertisements and promotional content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

pil

-0.67

Pag

-0.65

 perm

-0.64

 manned

-0.63

 arche

-0.62

 Priest

-0.62

rog

-0.61

 Shack

-0.61

 speeding

-0.61

 craw

-0.60

POSITIVE LOGITS

window

0.86

00200000

0.85

Anonymous

0.82

âĸ¬

0.81

TY

0.81

ãĥ³ãĤ¸

0.80

ategories

0.79

MSN

0.79

ãĥĥãĤ¯

0.79

âĵĺ

0.78

Activations Density 0.195%

abbreviated forms of words or phrases signifying emphasis or importance

references to online advertisements and promotional content

No Comments

No Known Activations