INDEX

Explanations

alphanumeric sequences mixed with abbreviations or acronyms

oai_token-act-pair · gpt-3.5-turbo

abbreviations or acronyms related to organizations or entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 puff

-0.70

cir

-0.67

Peb

-0.66

coc

-0.64

 Elixir

-0.61

 Mexicans

-0.61

ado

-0.60

 Scrib

-0.60

 Gree

-0.59

 Lear

-0.58

POSITIVE LOGITS

IUM

0.96

YD

0.95

IFE

0.95

WD

0.93

dit

0.93

IRE

0.92

YE

0.92

IR

0.90

HE

0.90

KI

0.90

Activations Density 0.142%

alphanumeric sequences mixed with abbreviations or acronyms

abbreviations or acronyms related to organizations or entities

No Comments

No Known Activations