INDEX

Explanations

the word "even"

oai_token-act-pair · gpt-3.5-turbo

the word "even."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 head

-0.69

 caps

-0.68

cap

-0.66

 detail

-0.64

 upside

-0.60

 preparations

-0.60

 preparation

-0.60

 version

-0.59

 component

-0.59

 client

-0.58

POSITIVE LOGITS

even

3.66

 EVEN

1.98

especially

1.83

despite

1.72

 even

1.62

Even

1.54

particularly

1.53

much

1.43

except

1.42

albeit

1.41

Activations Density 0.006%

the word "even"

the word "even."

No Comments

No Known Activations