INDEX

Explanations

dates presented in a particular format

oai_token-act-pair · gpt-3.5-turbo

sentence-ending punctuation marks, specifically periods

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 tremend

-0.76

 unpredict

-0.76

lihood

-0.75

 expended

-0.70

 pudding

-0.69

 leash

-0.69

 undecided

-0.68

 transact

-0.68

 everywhere

-0.67

 constantly

-0.67

POSITIVE LOGITS

↵

1.14

↵↵

1.13

Interested

1.07

Writing

1.03

 Speaking

1.01

Speaking

1.00

 According

0.98

According

0.97

 Instead

0.95

 Specifically

0.94

Activations Density 0.214%

dates presented in a particular format

sentence-ending punctuation marks, specifically periods

No Comments

No Known Activations