INDEX

Explanations

occurrences of the word "Chapel" followed by a sequence of numbers

oai_token-act-pair · gpt-3.5-turbo

references to specific locations or entities, with a focus on the term "husk" and the location "Chapel Hill."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ILE

-0.85

 Feldman

-0.78

chwitz

-0.75

 baseless

-0.68

 leases

-0.66

RT

-0.65

SD

-0.64

 matt

-0.64

paio

-0.64

tz

-0.63

POSITIVE LOGITS

thro

2.48

 Chapel

2.00

hus

1.96

hoe

1.73

throp

1.45

thus

1.31

Cho

1.20

 chapel

1.13

ourage

1.11

 Thro

1.06

Activations Density 0.053%

occurrences of the word "Chapel" followed by a sequence of numbers

references to specific locations or entities, with a focus on the term "husk" and the location "Chapel Hill."

No Comments

No Known Activations