INDEX

Explanations

imperative verbs instructing action

oai_token-act-pair · gpt-3.5-turbo

imperative verbs that suggest action or requests

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

DERR

-0.68

emale

-0.60

SUP

-0.59

CHR

-0.57

ropolis

-0.56

 polic

-0.56

prototype

-0.55

.","

-0.54

"},"

-0.54

irst

-0.53

POSITIVE LOGITS

 yourselves

0.95

 aside

0.79

 yourself

0.75

quote

0.74

mittedly

0.70

fax

0.69

 kidding

0.68

ichever

0.68

ably

0.67

esides

0.67

Activations Density 0.221%

imperative verbs instructing action

imperative verbs that suggest action or requests

No Comments

No Known Activations