INDEX

Explanations

instructions or commands telling someone what not to do

oai_token-act-pair · gpt-3.5-turbo

negative imperative statements or warnings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ciation

-0.78

quickShipAvailable

-0.74

¶æ

-0.72

 Reborn

-0.71

 proving

-0.67

GoldMagikarp

-0.66

arity

-0.64

omen

-0.63

 Liberation

-0.63

 Relief

-0.62

POSITIVE LOGITS

 underestimate

1.35

 confuse

1.23

 hesitate

1.14

 forget

1.09

 interfere

1.06

 misuse

1.04

 overe

1.03

 blindly

1.02

 bother

1.01

 EVER

1.01

Activations Density 0.129%

instructions or commands telling someone what not to do

negative imperative statements or warnings

No Comments

No Known Activations