INDEX

Explanations

questions with numbers or amounts in them

oai_token-act-pair · gpt-3.5-turbo

numerical references related to duration and frequency

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

roth

-0.59

pex

-0.57

ELL

-0.57

 drafting

-0.55

 chatter

-0.52

 buildup

-0.52

iller

-0.52

OA

-0.52

ahime

-0.51

oka

-0.51

POSITIVE LOGITS

 times

1.01

 nights

0.96

fold

0.93

 mornings

0.91

uously

0.89

 Nights

0.86

aciously

0.85

 weeks

0.82

 days

0.82

 months

0.79

Activations Density 0.132%

questions with numbers or amounts in them

numerical references related to duration and frequency

No Comments

No Known Activations