INDEX

Explanations

words that are commonly or typically associated with a certain context or action

oai_token-act-pair · gpt-3.5-turbo

the frequency of the word "usually."

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

 Further

-0.65

Fund

-0.64

asing

-0.57

bold

-0.56

Vote

-0.56

Merit

-0.55

iers

-0.54

ueller

-0.54

 future

-0.52

umbs

-0.51

POSITIVE LOGITS

 usually

3.27

 typically

2.71

usually

2.69

 normally

2.41

 often

2.29

typically

2.23

 invariably

2.19

often

2.15

 ordinarily

2.10

 Usually

2.09

Activations Density 0.012%

words that are commonly or typically associated with a certain context or action

the frequency of the word "usually."

No Comments

No Known Activations