INDEX

Explanations

political and legal terms or entities

oai_token-act-pair · gpt-3.5-turbo

references to government departments and legal proceedings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Indones

-0.70

 ancest

-0.70

Hop

-0.70

enes

-0.69

 Prelude

-0.68

 Proto

-0.68

 Roose

-0.68

acters

-0.67

Pure

-0.66

Portland

-0.65

POSITIVE LOGITS

 appointing

1.11

 appoint

1.10

 appointed

1.04

 investigating

0.99

 guidelines

0.97

 decree

0.97

 rulings

0.96

 overseeing

0.95

 reprim

0.94

 declass

0.93

Activations Density 1.074%

political and legal terms or entities

references to government departments and legal proceedings

No Comments

No Known Activations