INDEX

Explanations

analogies comparing one thing to another using words like "like" or "akin to" or "is like"

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

formance

-0.79

amily

-0.78

icipated

-0.75

vo

-0.71

merce

-0.71

 furthermore

-0.69

Site

-0.69

legates

-0.68

icip

-0.68

dor

-0.68

POSITIVE LOGITS

 aspirin

1.04

 Titanic

0.96

0.88

 cigarettes

0.87

 steroids

0.87

 chess

0.86

 Hitler

0.84

 Napoleon

0.83

 typew

0.83

 candy

0.83

Activations Density 0.736%

analogies comparing one thing to another using words like "like" or "akin to" or "is like"

No Comments

No Known Activations