INDEX

Explanations

instances where the concept of being open or accessible is mentioned

oai_token-act-pair · gpt-3.5-turbo

instances of the word "open" 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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Witches

-0.66

iage

-0.65

 Archangel

-0.64

 Lambert

-0.64

ORGE

-0.63

Kenn

-0.62

ILL

-0.62

hov

-0.62

Avg

-0.61

 Vill

-0.60

POSITIVE LOGITS

open

1.03

 open

0.93

office

0.91

 openness

0.80

ãĥ¼ãĤ¯

0.79

quartered

0.74

ĵĺ

0.74

ahead

0.73

POSE

0.72

 antioxid

0.71

Activations Density 0.029%

instances where the concept of being open or accessible is mentioned

instances of the word "open" in various contexts

No Comments

No Known Activations