INDEX

Explanations

names of political figures

oai_token-act-pair · gpt-3.5-turbo

mentions of various governors

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

 notebooks

-0.79

onyms

-0.72

DragonMagazine

-0.69

usercontent

-0.69

ãĤ»

-0.67

 Invasion

-0.67

drivers

-0.67

zsche

-0.65

ODUCT

-0.63

 positional

-0.63

POSITIVE LOGITS

 Bryant

0.78

 Cyrus

0.78

 Rudolph

0.68

 Rend

0.67

 Thib

0.67

 Alvarez

0.67

gov

0.67

 Atkins

0.67

 Bennett

0.67

itol

0.67

Activations Density 0.074%

names of political figures

mentions of various governors

No Comments

No Known Activations