INDEX

Explanations

dates and numbers

oai_token-act-pair · gpt-3.5-turbo

dates and numerical values associated with them

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

cientious

-0.73

laim

-0.67

aminer

-0.67

esse

-0.65

intend

-0.64

rider

-0.64

 Watts

-0.63

endo

-0.62

ducers

-0.62

uchs

-0.62

POSITIVE LOGITS

Ø¯

0.72

argo

0.67

ADRA

0.64

 Kraken

0.63

;;;;;;;;;;;;

0.61

Tokens

0.61

ĳå£«

0.61

 coasts

0.59

isin

0.59

 drills

0.58

Activations Density 0.299%

dates and numbers

dates and numerical values associated with them

No Comments

No Known Activations