INDEX

Explanations

proper nouns/entities, particularly related to individuals

oai_token-act-pair · gpt-3.5-turbo

names of characters or individuals, particularly in the context of media or storytelling

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

matically

-0.86

plain

-0.83

brance

-0.79

behind

-0.79

çĦ

-0.69

mable

-0.69

İĭ

-0.67

gling

-0.65

acement

-0.64

points

-0.63

POSITIVE LOGITS

icio

1.18

ahu

0.80

chi

0.76

del

0.75

uador

0.74

Sul

0.73

heastern

0.72

aido

0.70

eties

0.69

ctic

0.69

Activations Density 0.023%

proper nouns/entities, particularly related to individuals

names of characters or individuals, particularly in the context of media or storytelling

No Comments

No Known Activations