INDEX

Explanations

instructions or commands related to handling equipment or avoiding trouble

oai_token-act-pair · gpt-3.5-turbo

warnings and precautions related to safety and risk

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 awaited

-0.81

orest

-0.81

 awaits

-0.77

 congratulations

-0.73

 grateful

-0.73

 thanked

-0.72

 thanking

-0.71

 thank

-0.67

thank

-0.65

 patiently

-0.65

POSITIVE LOGITS

 inappropriately

1.28

 unnecessarily

1.24

 overboard

1.23

 undue

1.23

 dangerously

1.22

 inadvertently

1.18

 interfere

1.16

 unintentionally

1.13

 excessively

1.13

 interfering

1.13

Activations Density 1.246%

instructions or commands related to handling equipment or avoiding trouble

warnings and precautions related to safety and risk

No Comments

No Known Activations