INDEX

Explanations

instances of altruistic actions or good deeds

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

redients

-0.78

 rough

-0.64

 preliminary

-0.63

 basics

-0.61

ellow

-0.57

 Rough

-0.56

ications

-0.55

 Extras

-0.55

 Upton

-0.55

ils

-0.53

POSITIVE LOGITS

never

2.09

 never

2.06

 NEVER

1.96

 Never

1.87

Never

1.79

 ALWAYS

1.69

 always

1.68

always

1.65

 ever

1.58

 Always

1.50

Activations Density 0.208%

instances of altruistic actions or good deeds

No Comments

No Known Activations