INDEX

Explanations

phrases related to unofficial or indirect authority, positions, or statuses

oai_token-act-pair · gpt-3.5-turbo

terms related to de facto situations or statuses

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

lyss

-0.72

Õ

-0.71

 Braz

-0.71

Course

-0.70

cule

-0.68

urgy

-0.68

nder

-0.67

rik

-0.67

gard

-0.67

rug

-0.67

POSITIVE LOGITS

 facto

1.04

 monarchy

0.80

 Lumpur

0.79

ensable

0.79

doms

0.78

 unofficial

0.74

 Dhabi

0.72

ignment

0.71

 unrecogn

0.71

ignty

0.70

Activations Density 0.016%

phrases related to unofficial or indirect authority, positions, or statuses

terms related to de facto situations or statuses

No Comments

No Known Activations