INDEX

Explanations

proper nouns starting with "Ad" followed by a high variety of words, suggesting a possible focus on a specific entity or concept

oai_token-act-pair · gpt-3.5-turbo

references to advertisements or promotional content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

Ń·

-0.78

OUP

-0.74

 Rouge

-0.71

 Ragnarok

-0.71

terday

-0.69

Jr

-0.69

KEY

-0.67

\\\\

-0.66

ãĥīãĥ©ãĤ´ãĥ³

-0.66

POSITIVE LOGITS

vertisements

1.27

vertis

1.25

olesc

1.15

olescent

1.11

ept

1.08

mittedly

1.05

hesive

1.02

verb

1.00

vertising

0.99

rien

0.98

Activations Density 0.008%

proper nouns starting with "Ad" followed by a high variety of words, suggesting a possible focus on a specific entity or concept

references to advertisements or promotional content

No Comments

No Known Activations