INDEX

Explanations

references to written or drawn content that has been altered or edited

oai_token-act-pair · gpt-3.5-turbo

words that have specific suffixes or endings related to actions or conditions

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

imar

-0.67

ð

-0.65

 Skydragon

-0.63

ategory

-0.63

bitious

-0.62

emetery

-0.62

iltr

-0.61

wings

-0.59

ritional

-0.59

isl

-0.58

POSITIVE LOGITS

lihood

0.97

Oo

0.75

Sov

0.75

çīĪ

0.72

 Dragonbound

0.72

 thereof

0.72

advertisement

0.70

 extraord

0.66

oute

0.64

 Scotia

0.64

Activations Density 0.211%

references to written or drawn content that has been altered or edited

words that have specific suffixes or endings related to actions or conditions

No Comments

No Known Activations