INDEX

Explanations

numbers representing quantities or quantities of actions

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "once" and variations indicating repetition or additional items

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

utra

-0.71

emet

-0.64

stood

-0.63

eve

-0.60

ĸļ

-0.58

obin

-0.58

owned

-0.58

haw

-0.58

onomy

-0.57

emed

-0.57

POSITIVE LOGITS

 more

1.86

 MORE

1.52

more

1.49

More

1.42

 further

1.36

 More

1.32

 longer

1.26

 additional

1.23

MORE

1.18

 fewer

1.13

Activations Density 0.598%

numbers representing quantities or quantities of actions

occurrences of the word "once" and variations indicating repetition or additional items

No Comments

No Known Activations