INDEX

Explanations

phrases indicating the need for correction or accuracy

oai_token-act-pair · gpt-3.5-turbo

mentions of correctness or accuracy in statements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

CHO

-0.82

EMOTE

-0.71

aden

-0.68

neys

-0.68

lust

-0.67

 Pharma

-0.67

rug

-0.64

 Sunset

-0.63

jit

-0.63

cheon

-0.62

POSITIVE LOGITS

correct

1.08

Correct

1.04

ives

0.91

 correct

0.90

 corrected

0.87

 Correct

0.86

orate

0.79

 spelling

0.78

 answer

0.78

 correctness

0.78

Activations Density 0.007%

phrases indicating the need for correction or accuracy

mentions of correctness or accuracy in statements

No Comments

No Known Activations