INDEX

Explanations

phrases related to headlines or headings

oai_token-act-pair · gpt-3.5-turbo

instances of the term "lines" in conjunction with numeric values

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

 Mushroom

-0.67

 mater

-0.66

 bought

-0.65

 confer

-0.64

 purchased

-0.62

 proceed

-0.61

mug

-0.59

 misc

-0.59

 suspect

-0.59

 purpose

-0.58

POSITIVE LOGITS

lines

4.54

line

3.08

lined

2.44

lining

2.23

LINE

2.14

 lines

2.09

liners

2.05

 Lines

1.99

Line

1.87

liner

1.84

Activations Density 0.007%

phrases related to headlines or headings

instances of the term "lines" in conjunction with numeric values

No Comments

No Known Activations