INDEX

Explanations

statements or commands

oai_token-act-pair · gpt-3.5-turbo

elements of confusion or miscommunication in conversations

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

uador

-0.67

 ageing

-0.66

 citizenship

-0.62

rament

-0.62

cius

-0.61

 aging

-0.61

urtle

-0.60

awei

-0.60

ijing

-0.60

heit

-0.58

POSITIVE LOGITS

Seriously

0.69

Especially

0.69

âĶĢâĶĢâĶĢâĶĢ

0.65

 Norn

0.64

 Dare

0.64

Lol

0.64

Quotes

0.63

????????

0.63

ÂłÂłÂł

0.63

 Costs

0.63

Activations Density 0.309%

statements or commands

elements of confusion or miscommunication in conversations

No Comments

No Known Activations