INDEX

Explanations

dates or years

oai_token-act-pair · gpt-3.5-turbo

references to the year 2017

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

tub

-0.69

 repro

-0.65

imore

-0.62

cipl

-0.61

 throb

-0.61

cephal

-0.61

anguage

-0.60

 breathing

-0.58

onic

-0.58

 polyg

-0.58

POSITIVE LOGITS

âĢ²

0.81

halla

0.80

å¹

0.79

âĢ³

0.73

0.72

 Winners

0.70

Ä±

0.69

bucks

0.69

ESV

0.69

 onwards

0.67

Activations Density 0.032%

dates or years

references to the year 2017

No Comments

No Known Activations