INDEX

Explanations

phrases indicating user interaction and instructions

oai_token-act-pair · gpt-3.5-turbo

future-oriented phrases expressing intent or necessity

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

 Sandwich

-0.68

ONSORED

-0.63

states

-0.60

 Tsukuyomi

-0.59

ihad

-0.58

ahead

-0.58

MO

-0.57

blooded

-0.57

 Fourth

-0.57

farious

-0.56

POSITIVE LOGITS

 need

1.18

 find

1.15

 notice

1.12

 encounter

1.12

 hear

1.09

 discover

1.08

 receive

1.07

see

1.07

 undoubtedly

1.01

be

0.99

Activations Density 0.114%

phrases indicating user interaction and instructions

future-oriented phrases expressing intent or necessity

No Comments

No Known Activations