INDEX

Explanations

written text fragments

oai_token-act-pair · gpt-3.5-turbo

references to works that have been written

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

nel

-0.81

nels

-0.80

Ĭ±

-0.79

Ĥª

-0.77

Sensor

-0.72

etz

-0.69

 Tide

-0.68

ĪĴ

-0.68

olate

-0.67

abol

-0.66

POSITIVE LOGITS

 aloud

0.93

escription

0.80

acters

0.78

 down

0.72

itatively

0.70

 expressly

0.70

 comprehension

0.70

by

0.69

 representations

0.69

 prescriptions

0.68

Activations Density 0.040%

written text fragments

references to works that have been written

No Comments

No Known Activations