INDEX

Explanations

proper nouns, possibly related to social media profiles or blog posts

oai_token-act-pair · gpt-3.5-turbo

names, identifiers, or titles related to individuals and their works

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

hement

-0.89

apon

-0.67

igi

-0.66

uca

-0.64

ouble

-0.63

 XIII

-0.62

cius

-0.61

lio

-0.61

roid

-0.61

uto

-0.59

POSITIVE LOGITS

ÂłÂłÂłÂłÂłÂłÂłÂł

0.80

ÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂł

0.78

ÂłÂłÂł

0.76

ÂłÂł

0.70

Posted

0.67

Filename

0.67

ÂłÂłÂłÂł

0.66

Synopsis

0.64

 Codes

0.64

oding

0.58

Activations Density 0.358%

proper nouns, possibly related to social media profiles or blog posts

names, identifiers, or titles related to individuals and their works

No Comments

No Known Activations