INDEX

Explanations

phrases related to historical events or programs

oai_token-act-pair · gpt-3.5-turbo

phrases that include the word "under" indicating subordination or categorization within a context

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ern

-0.76

eka

-0.70

ibia

-0.68

Inside

-0.65

ument

-0.65

veland

-0.65

iva

-0.65

ades

-0.63

izu

-0.60

\-

-0.60

POSITIVE LOGITS

pins

1.11

lines

1.09

cuts

1.05

 supervision

1.04

cutting

1.03

 advis

1.02

lie

0.99

lining

0.98

written

0.96

neath

0.95

Activations Density 0.058%

phrases related to historical events or programs

phrases that include the word "under" indicating subordination or categorization within a context

No Comments

No Known Activations