INDEX

Explanations

references to robots

oai_token-act-pair · gpt-3.5-turbo

references to robots

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ingle

-0.84

 holder

-0.74

iece

-0.74

 Yorker

-0.71

pal

-0.71

opal

-0.68

isal

-0.68

Ã©t

-0.67

nah

-0.67

 Evening

-0.67

POSITIVE LOGITS

 robots

3.74

 Robots

2.81

 bots

2.46

 robotics

2.19

 robot

2.06

bots

2.00

 drones

1.97

 Robotics

1.93

 machines

1.91

 robotic

1.75

Activations Density 0.017%

references to robots

references to robots

No Comments

No Known Activations