INDEX

Explanations

phrases or contexts where something is explicitly stated or emphasized

oai_token-act-pair · gpt-3.5-turbo

instances of the word "pronounced" used in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 items

-0.64

 Inqu

-0.62

buck

-0.60

share

-0.60

lists

-0.59

acles

-0.58

 item

-0.58

media

-0.56

cover

-0.56

 supplement

-0.56

POSITIVE LOGITS

 pronounced

4.03

pron

1.25

 noticeable

1.16

 subdued

1.10

 spelled

1.09

 pron

1.04

 proclaimed

0.91

 marked

0.91

 acute

0.89

 predominant

0.88

Activations Density 0.009%

phrases or contexts where something is explicitly stated or emphasized

instances of the word "pronounced" used in various contexts

No Comments

No Known Activations