INDEX

Explanations

names of people or entities related to various activities

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

references to the name "Graeme."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_post

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

 intervals

-0.64

NPR

-0.64

 noses

-0.62

 politics

-0.60

 misinformation

-0.59

 resemb

-0.59

 papers

-0.59

 incidentally

-0.58

 false

-0.58

 last

-0.57

POSITIVE LOGITS

ffiti

1.87

eme

1.43

veyard

1.35

eco

1.25

ffen

1.14

ppa

1.13

inger

1.12

flex

1.11

veland

1.11

vel

1.10

Activations Density 0.022%

names of people or entities related to various activities

references to the name "Graeme."

No Comments

No Known Activations