INDEX

Explanations

phrases related to providing or granting something to someone

oai_token-act-pair · gpt-3.5-turbo

terms related to the concept of protection or privileges being granted

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

so

-0.70

can

-0.70

patch

-0.68

 tune

-0.67

cum

-0.65

opers

-0.65

atro

-0.64

chem

-0.64

link

-0.63

analysis

-0.63

POSITIVE LOGITS

 afforded

1.23

ĸļ

1.12

utical

0.99

ãĥ¼ãĥĨ

0.98

xual

0.93

swer

0.91

ufact

0.89

 Granted

0.84

enance

0.84

ividual

0.80

Activations Density 0.011%

phrases related to providing or granting something to someone

terms related to the concept of protection or privileges being granted

No Comments

No Known Activations