INDEX

Explanations

instructions on how to perform specific tasks or procedures

oai_token-act-pair · gpt-3.5-turbo

second-person pronouns and directives addressing the reader

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

Georg

-0.59

 millenn

-0.54

REDACTED

-0.54

ãĤ´ãĥ³

-0.54

 earthqu

-0.53

adium

-0.53

amaz

-0.52

hearted

-0.52

 Atlantis

-0.52

Joined

-0.52

POSITIVE LOGITS

'll

1.27

 shouldn

1.20

 MUST

1.16

're

1.15

 need

1.15

 should

1.11

 SHOULD

1.10

 want

1.08

 NEED

1.08

 might

1.06

Activations Density 0.161%

instructions on how to perform specific tasks or procedures

second-person pronouns and directives addressing the reader

No Comments

No Known Activations