INDEX

Explanations

mentions of a specific place or organization, 'CH'

oai_token-act-pair · gpt-3.5-turbo

mentions of a specific television channel

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĥ¼ãĥĨãĤ£

-1.03

gaard

-0.88

berman

-0.85

ãĤ´ãĥ³

-0.83

glas

-0.79

 Murdoch

-0.75

ãĥ¼ãĥĨ

-0.75

ãĥ¼ãĥ³

-0.69

Reloaded

-0.69

sov

-0.68

POSITIVE LOGITS

APTER

1.23

ECK

1.04

bones

0.99

bone

0.94

ivalry

0.93

olester

0.83

aos

0.82

apeake

0.81

ieft

0.79

arest

0.77

Activations Density 0.006%

mentions of a specific place or organization, 'CH'

mentions of a specific television channel

No Comments

No Known Activations