INDEX

Explanations

phrases related to specific locations such as streets and blocks

oai_token-act-pair · gpt-3.5-turbo

references to blocks and community-related terms in urban contexts

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

verty

-0.69

STEM

-0.69

ahime

-0.68

ndra

-0.65

 Patron

-0.62

 orient

-0.62

 leaps

-0.62

uld

-0.62

 Saban

-0.61

 lull

-0.60

POSITIVE LOGITS

 flats

0.83

airs

0.75

 highways

0.69

 checkpoints

0.68

ctive

0.67

 Entry

0.66

ading

0.65

google

0.61

 bridges

0.60

 attempts

0.60

Activations Density 0.242%

phrases related to specific locations such as streets and blocks

references to blocks and community-related terms in urban contexts

No Comments

No Known Activations