INDEX

Explanations

specific mentions of the word "users" in various contexts

oai_token-act-pair · gpt-3.5-turbo

mentions of users in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

western

-0.69

 Hurricanes

-0.67

 Wolver

-0.65

 Shore

-0.64

FLAG

-0.64

Rib

-0.62

 Baptist

-0.62

ttp

-0.62

erella

-0.60

 Corridor

-0.59

POSITIVE LOGITS

hip

1.08

pace

1.02

paces

0.91

cript

0.88

ettings

0.83

 interfaces

0.82

mens

0.80

 interface

0.78

criptions

0.78

hare

0.76

Activations Density 0.028%

specific mentions of the word "users" in various contexts

mentions of users in various contexts

No Comments

No Known Activations