INDEX

Explanations

instances where there is a contrast or unexpected outcome compared to a previous state or expectation

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

insula

-0.79

ASE

-0.66

 Stim

-0.62

 elimination

-0.58

Winter

-0.58

idium

-0.58

ufact

-0.58

Delivery

-0.57

 Correction

-0.57

anooga

-0.56

POSITIVE LOGITS

birth

1.08

 retains

0.91

 manages

0.88

born

0.80

 managed

0.79

 exists

0.78

 haun

0.76

 maintains

0.75

 manage

0.74

 intact

0.73

Activations Density 0.052%

instances where there is a contrast or unexpected outcome compared to a previous state or expectation

No Comments

No Known Activations