INDEX

Explanations

terms related to different types of progress or advancements

oai_token-act-pair · gpt-3.5-turbo

references to significant developments or events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

asus

-0.79

otto

-0.77

unts

-0.75

mi

-0.75

cer

-0.74

amina

-0.73

lua

-0.69

heed

-0.68

vol

-0.67

que

-0.66

POSITIVE LOGITS

 developments

1.37

hower

0.92

 horizont

0.88

afety

0.82

 announcements

0.80

cape

0.79

 trends

0.78

 setbacks

0.75

 directives

0.74

 advancements

0.73

Activations Density 0.004%

terms related to different types of progress or advancements

references to significant developments or events

No Comments

No Known Activations