INDEX

Explanations

words related to lines or linear objects

oai_token-act-pair · gpt-3.5-turbo

references to "line" 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

 vulner

-0.64

 hiber

-0.60

 incub

-0.59

 Ukrain

-0.59

 mashed

-0.58

 Kitt

-0.57

à¼

-0.56

 sake

-0.55

 simulac

-0.55

sir

-0.55

POSITIVE LOGITS

line

4.40

lines

3.44

LINE

2.65

lining

2.42

lined

2.38

liner

2.37

Line

2.36

liners

2.22

 line

2.14

 Line

1.87

Activations Density 0.007%

words related to lines or linear objects

references to "line" in various contexts

No Comments

No Known Activations