INDEX

Explanations

words related to a specific topic or theme, specifically phrases containing "related to."

oai_token-act-pair · gpt-3.5-turbo

references to connections or associations with various topics

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

anse

-0.75

oise

-0.70

ossus

-0.68

olls

-0.67

$$$$

-0.64

Peb

-0.63

maximum

-0.62

undai

-0.61

awks

-0.61

oliath

-0.59

POSITIVE LOGITS

 related

3.70

 Related

2.30

related

2.25

 unrelated

1.91

 associated

1.85

 relating

1.79

 relates

1.56

 pertaining

1.52

 linked

1.51

 relate

1.49

Activations Density 0.010%

words related to a specific topic or theme, specifically phrases containing "related to."

references to connections or associations with various topics

No Comments

No Known Activations