INDEX

Explanations

statements related to an increase in quantity or intensity

oai_token-act-pair · gpt-3.5-turbo

instances of the word "increased"

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

Bio

-0.78

puted

-0.74

 Impossible

-0.74

 Tycoon

-0.72

 Mole

-0.72

verse

-0.70

 Geneva

-0.68

isters

-0.68

 Outs

-0.68

heim

-0.67

POSITIVE LOGITS

xual

1.04

 subsequ

1.04

 tremend

1.01

 confir

0.96

 millenn

0.94

 awareness

0.93

 unintention

0.93

 eleph

0.92

 responsiveness

0.91

 carbohyd

0.90

Activations Density 0.019%

statements related to an increase in quantity or intensity

instances of the word "increased"

No Comments

No Known Activations