INDEX

Explanations

comparative phrases expressing improvement or preference

oai_token-act-pair · gpt-3.5-turbo

phrases that express the concept of improvement or enhancement

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Dresden

-0.67

aly

-0.67

 Pione

-0.64

idon

-0.63

heter

-0.63

mberg

-0.62

anuts

-0.61

ategory

-0.61

 Canaver

-0.61

leted

-0.60

POSITIVE LOGITS

 than

1.17

 suited

1.06

 behaved

0.95

seller

0.90

 Than

0.84

than

0.82

 acquainted

0.80

 quality

0.78

 vers

0.75

iating

0.75

Activations Density 0.039%

comparative phrases expressing improvement or preference

phrases that express the concept of improvement or enhancement

No Comments

No Known Activations