INDEX

Explanations

the word "orders" with a particular emphasis on occurrences when it is activated with a high level of intensity

oai_token-act-pair · gpt-3.5-turbo

references to "orders" or directives 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 @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Miliband

-0.72

ãĤ¨

-0.70

vae

-0.70

pmwiki

-0.68

Ott

-0.68

 Dull

-0.67

Marginal

-0.66

ES

-0.66

 Libre

-0.66

de

-0.65

POSITIVE LOGITS

 orders

1.41

 ordering

1.06

avorite

0.96

etary

0.96

 Orders

0.95

hops

0.94

 order

0.89

 ordered

0.89

ylum

0.88

lies

0.87

Activations Density 0.012%

the word "orders" with a particular emphasis on occurrences when it is activated with a high level of intensity

references to "orders" or directives in various contexts

No Comments

No Known Activations