INDEX

Explanations

emphasis or curiosity expressed through the word "what"

oai_token-act-pair · gpt-3.5-turbo

interrogative phrases or words

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

uling

-0.68

psc

-0.66

uating

-0.63

ean

-0.62

ulkan

-0.61

istani

-0.61

Ń·

-0.60

mens

-0.59

atis

-0.59

POR

-0.59

POSITIVE LOGITS

soever

1.35

 else

1.01

 happened

0.99

 ensued

0.95

 happens

0.94

 transpired

0.91

 better

0.89

 follows

0.81

?!

0.80

 separates

0.80

Activations Density 0.076%

emphasis or curiosity expressed through the word "what"

interrogative phrases or words

No Comments

No Known Activations