INDEX

Explanations

names related to the term "Kan" followed by a digit (likely a specific identifier or code)

oai_token-act-pair · gpt-3.5-turbo

mentions of the word "Kan" 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 @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ears

-0.82

esp

-0.69

ought

-0.65

FORE

-0.65

fore

-0.63

ista

-0.61

ear

-0.61

ther

-0.61

\\\\

-0.60

yrim

-0.59

POSITIVE LOGITS

Kan

3.87

kan

2.03

Kaw

1.37

 Karn

1.36

kan

1.33

KS

1.32

Kab

1.30

Kot

1.24

Kou

1.23

 Kahn

1.22

Activations Density 0.015%

names related to the term "Kan" followed by a digit (likely a specific identifier or code)

mentions of the word "Kan" in various contexts

No Comments

No Known Activations