INDEX

Explanations

additions, new elements, or layers

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

stood

-0.66

open

-0.62

published

-0.61

Present

-0.58

Against

-0.57

acebook

-0.56

CBC

-0.56

doi

-0.56

Win

-0.56

POSITIVE LOGITS

 thereto

1.04

endum

1.04

 flair

0.95

 layer

0.90

 dimension

0.89

 extra

0.88

 onto

0.87

 insult

0.84

 suffix

0.83

 additional

0.82

Activations Density 0.176%

additions, new elements, or layers

No Comments

No Known Activations