INDEX

Explanations

instructions or steps related to physical actions or tasks

oai_token-act-pair · gpt-3.5-turbo

instructions or steps related to processes or techniques

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

 tariff

-0.83

vou

-0.80

ylum

-0.77

 tariffs

-0.74

 boycot

-0.73

 categ

-0.72

 enrolled

-0.72

 informed

-0.72

 denounced

-0.72

 hosp

-0.70

POSITIVE LOGITS

Remove

1.11

 Remove

1.07

 Slide

1.07

Repeat

1.06

Use

1.03

 Notice

1.01

Pull

0.99

Insert

0.99

Use

0.99

 Apply

0.99

Activations Density 0.342%

instructions or steps related to physical actions or tasks

instructions or steps related to processes or techniques

No Comments

No Known Activations