INDEX

Explanations

terms related to the substance "tar"

oai_token-act-pair · gpt-3.5-turbo

references to tar

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

etts

-0.78

assetsadobe

-0.75

rontal

-0.73

egg

-0.72

ESSION

-0.71

 Startup

-0.69

sense

-0.68

shire

-0.67

ETF

-0.67

swers

-0.67

POSITIVE LOGITS

tar

1.37

 sands

1.12

 corrid

0.82

ãĥ£

0.79

anium

0.79

 blister

0.75

balls

0.72

 toile

0.72

 psychiat

0.70

 puff

0.69

Activations Density 0.006%

terms related to the substance "tar"

references to tar

No Comments

No Known Activations