INDEX

Explanations

patterns or sequences

oai_token-act-pair · gpt-3.5-turbo

repeated references to visual elements and attributes in descriptions

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

ŃĶ

-0.86

yss

-0.70

enum

-0.70

yrinth

-0.70

ressor

-0.68

bid

-0.67

 Stronghold

-0.67

aternity

-0.66

ntil

-0.63

 discipl

-0.62

POSITIVE LOGITS

mith

1.14

hips

1.13

ynthesis

1.10

heet

1.09

pots

1.08

peed

0.99

hops

0.97

pace

0.96

poons

0.96

paces

0.96

Activations Density 0.019%

patterns or sequences

repeated references to visual elements and attributes in descriptions

No Comments

No Known Activations