INDEX

Explanations

verbs related to utilizing something or performing an action

oai_token-act-pair · gpt-3.5-turbo

instances of the word "use" in various contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Ń·

-0.68

RANT

-0.66

SPONSORED

-0.65

cffffcc

-0.64

 Empires

-0.63

 Barg

-0.63

uberty

-0.62

 Griff

-0.61

 behold

-0.61

mother

-0.61

POSITIVE LOGITS

fully

1.02

FUL

0.96

ancies

0.96

fulness

0.86

iences

0.85

ience

0.81

full

0.78

 shortcuts

0.77

ibles

0.76

 aliases

0.76

Activations Density 0.108%

verbs related to utilizing something or performing an action

instances of the word "use" in various contexts

No Comments

No Known Activations