INDEX

Explanations

references to religious entities, concepts, or activities

oai_token-act-pair · gpt-3.5-turbo

words associated with religion and its influence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 fixes

-0.69

 reset

-0.68

fix

-0.68

 plan

-0.65

 plans

-0.62

 rede

-0.61

 recap

-0.61

 mask

-0.60

fixes

-0.60

 modules

-0.60

POSITIVE LOGITS

igious

4.50

igion

2.61

inent

1.73

entious

1.27

inence

1.21

ige

1.17

igators

1.09

igent

1.06

igation

1.04

imental

1.02

Activations Density 0.020%

references to religious entities, concepts, or activities

words associated with religion and its influence

No Comments

No Known Activations