INDEX

Explanations

phrases related to evaluation or exploration

oai_token-act-pair · gpt-3.5-turbo

phrases indicating objectives or methods for acquiring knowledge and understanding

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ãĥĲ

-0.66

 squares

-0.64

 towels

-0.62

 feces

-0.60

 benches

-0.58

 drowned

-0.58

 veins

-0.58

 Croat

-0.57

ãģ®ç

-0.56

 peas

-0.56

POSITIVE LOGITS

INESS

0.69

iona

0.68

ocious

0.66

uce

0.65

 Higher

0.65

umar

0.65

ocard

0.63

ocation

0.63

FK

0.63

 Ratio

0.62

Activations Density 0.336%

phrases related to evaluation or exploration

phrases indicating objectives or methods for acquiring knowledge and understanding

No Comments

No Known Activations