INDEX

Explanations

The main thing this neuron does is find events where a specific location becomes the first or one of the first to do something

oai_token-act-pair · gpt-3.5-turbo

instances of countries being mentioned in relation to specific statuses or rankings

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

åħī

-0.64

worm

-0.64

mediated

-0.63

 Peng

-0.62

dfx

-0.62

fram

-0.61

matter

-0.61

cv

-0.60

 illusions

-0.60

POSITIVE LOGITS

 safest

0.97

 destination

0.89

 birthplace

0.87

swing

0.87

 destinations

0.83

 havens

0.83

 battleground

0.81

 capitals

0.78

 jurisdictions

0.74

orest

0.74

Activations Density 0.186%

The main thing this neuron does is find events where a specific location becomes the first or one of the first to do something

instances of countries being mentioned in relation to specific statuses or rankings

No Comments

No Known Activations