INDEX

Explanations

proper names of people or possibly locations, as well as locations

oai_token-act-pair · gpt-3.5-turbo

proper nouns, specifically names and surnames

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

represented

-0.75

patient

-0.73

Fed

-0.69

âĵĺ

-0.65

posed

-0.64

stood

-0.59

ataka

-0.59

 stubborn

-0.58

abol

-0.58

CLAIM

-0.58

POSITIVE LOGITS

 IMAGES

1.05

(@

1.02

 Photography

1.01

Jr

0.97

 Random

0.96

via

0.88

 Courtesy

0.87

baum

0.83

Sr

0.81

Activations Density 0.126%

proper names of people or possibly locations, as well as locations

proper nouns, specifically names and surnames

No Comments

No Known Activations