INDEX

Explanations

mentions or references to the country India

oai_token-act-pair · gpt-3.5-turbo

instances of the word "India."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

yright

-0.97

umbnails

-0.91

htaking

-0.82

olitan

-0.81

esters

-0.81

umbn

-0.80

cies

-0.79

ynt

-0.77

iliary

-0.77

ients

-0.76

POSITIVE LOGITS

India

1.12

 Pradesh

1.09

 India

1.08

 Sharma

1.06

 Pakistan

1.01

Pakistan

1.01

 Hindus

1.00

 Modi

0.98

 Kashmir

0.94

 Punjab

0.92

Activations Density 0.017%

mentions or references to the country India

instances of the word "India."

No Comments

No Known Activations