INDEX

Explanations

names of political figures

oai_token-act-pair · gpt-3.5-turbo

references to U.S. senators and their activities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 nature

-0.67

 appraisal

-0.66

 managerial

-0.65

 apart

-0.65

 underpin

-0.64

 behavi

-0.64

ussion

-0.63

zb

-0.63

 spac

-0.63

erie

-0.63

POSITIVE LOGITS

 Dianne

1.17

 Lindsey

1.09

 Cory

1.05

Ted

1.03

 Barack

1.02

 Bernie

1.02

 Grassley

1.02

 Jeanne

1.02

 Kirst

1.00

 Chuck

0.99

Activations Density 0.044%

names of political figures

references to U.S. senators and their activities

No Comments

No Known Activations