INDEX

Explanations

phrases indicating approval or agreement

oai_token-act-pair · gpt-3.5-turbo

conversational expressions and phrases commonly used in casual dialogue

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

-0.89

imum

-0.77

ombat

-0.74

ileged

-0.72

Mand

-0.71

leading

-0.69

¿½

-0.68

imon

-0.67

İĭ

-0.67

isively

-0.67

POSITIVE LOGITS

 forgot

0.95

 kinda

0.95

 sucks

0.85

 totally

0.79

 kidding

0.79

 goof

0.79

 tricked

0.73

 cheated

0.72

 fucked

0.71

 typo

0.71

Activations Density 0.263%

phrases indicating approval or agreement

conversational expressions and phrases commonly used in casual dialogue

No Comments

No Known Activations