INDEX

Explanations

phrases implying desire, preference, or approval

oai_token-act-pair · gpt-3.5-turbo

expressions of audience desires and preferences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ãĥķ

-0.76

 teasp

-0.75

çĶ°

-0.69

Aim

-0.67

adem

-0.62

wu

-0.61

Cooldown

-0.60

 suscept

-0.59

 workshop

-0.59

irlfriend

-0.59

POSITIVE LOGITS

 themselves

0.89

 their

0.88

 THEIR

0.71

 choices

0.70

their

0.68

 certainty

0.67

 these

0.67

 knowing

0.66

 seeing

0.66

 predictable

0.66

Activations Density 0.407%

phrases implying desire, preference, or approval

expressions of audience desires and preferences

No Comments

No Known Activations