INDEX

Explanations

words related to certifications or credentials

oai_token-act-pair · gpt-3.5-turbo

specific letters or characters within a structured context, like programming or markup languages

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 eleph

-0.76

 metic

-0.75

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.68

Þ

-0.68

ß

-0.67

 redes

-0.66

 conclud

-0.66

 proport

-0.64

ò

-0.64

 enthusi

-0.63

POSITIVE LOGITS

arma

0.87

aper

0.86

apper

0.84

ork

0.84

igger

0.81

irc

0.80

unc

0.80

undle

0.80

aps

0.79

anc

0.79

Activations Density 0.110%

words related to certifications or credentials

specific letters or characters within a structured context, like programming or markup languages

No Comments

No Known Activations