INDEX

Explanations

words indicating desire or intent

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

soever

-0.66

 forbids

-0.63

quished

-0.63

Reader

-0.61

mining

-0.60

audio

-0.60

oused

-0.60

ridden

-0.59

 Norton

-0.59

Cir

-0.58

POSITIVE LOGITS

 succeed

1.38

 behave

1.22

be

1.15

 realise

1.14

 realize

1.12

 acknowledge

1.04

 know

1.04

 come

1.04

 thrive

1.03

 adhere

1.03

Activations Density 0.126%

words indicating desire or intent

No Comments

No Known Activations