INDEX

Explanations

phrases related to well-known facts or information

oai_token-act-pair · gpt-3.5-turbo

repeated patterns and phrases that indicate commonality or prevalence

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

culus

-0.80

nings

-0.69

 Recon

-0.67

ucker

-0.67

 punches

-0.62

 Sick

-0.62

 stuffing

-0.61

 Tycoon

-0.61

 Sadd

-0.60

sis

-0.59

POSITIVE LOGITS

 debated

0.87

recated

0.87

 ubiqu

0.85

inventoryQuantity

0.84

ocry

0.84

 unch

0.83

 prolifer

0.81

 internationally

0.79

 ubiquitous

0.79

accessible

0.78

Activations Density 0.438%

phrases related to well-known facts or information

repeated patterns and phrases that indicate commonality or prevalence

No Comments

No Known Activations