INDEX

Explanations

phrases related to news articles or reports

oai_token-act-pair · gpt-3.5-turbo

phrases related to media reports and inaccuracies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

Cub

-1.00

zona

-0.88

Weiss

-0.88

EStream

-0.84

aeda

-0.84

ella

-0.80

++)

-0.77

podcast

-0.73

ktop

-0.72

[+]

-0.72

POSITIVE LOGITS

 Kejriwal

1.36

 Uttar

1.32

BJP

1.30

 Kumar

1.29

 Maharashtra

1.28

 Shiv

1.27

 Punjab

1.27

 Rahul

1.26

 Bihar

1.24

 Gujarat

1.23

Activations Density 0.290%

phrases related to news articles or reports

phrases related to media reports and inaccuracies

No Comments

No Known Activations