INDEX

Explanations

phrases where the word "Never" is strongly emphasized

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "never" or "always."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ipation

-0.75

Els

-0.72

åĤ

-0.71

CRIP

-0.67

lass

-0.66

 Burk

-0.64

 Beir

-0.64

 Austral

-0.64

lycer

-0.62

uers

-0.62

POSITIVE LOGITS

theless

1.28

entimes

0.83

 underestimate

0.80

 forgot

0.78

 listened

0.77

winter

0.76

 Asked

0.76

 Detected

0.76

 hesitated

0.75

Never

0.75

Activations Density 0.008%

phrases where the word "Never" is strongly emphasized

references to the concept of "never" or "always."

No Comments

No Known Activations