INDEX

Explanations

proper names and titles, particularly related to names starting with "Carl" and "Karl"

oai_token-act-pair · gpt-3.5-turbo

references to individuals, particularly those with the name "Carl" or similar variations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 concess

-0.76

 includ

-0.71

 overdue

-0.66

jri

-0.66

 DRAG

-0.66

theless

-0.65

dinand

-0.64

 recip

-0.64

 constitu

-0.63

 Reincarn

-0.62

POSITIVE LOGITS

å°Ĩ

0.78

ãĤ¼ãĤ¦ãĤ¹

0.78

stad

0.75

puff

0.74

ãĥ¢

0.73

ãĥ¡

0.72

ocker

0.69

ãĥ¼ãĥĨ

0.69

eston

0.69

Neh

0.69

Activations Density 0.206%

proper names and titles, particularly related to names starting with "Carl" and "Karl"

references to individuals, particularly those with the name "Carl" or similar variations

No Comments

No Known Activations