INDEX

Explanations

verbs related to interaction or conflict

oai_token-act-pair · gpt-3.5-turbo

instances of the word "engage" 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 @ 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

IFrame

Link

Negative Logits

ray

-0.74

 Preservation

-0.71

 Virus

-0.68

 Rita

-0.67

 pale

-0.65

 Param

-0.65

 rays

-0.62

Ain

-0.62

Rum

-0.62

grain

-0.61

POSITIVE LOGITS

 engage

3.34

 engages

2.69

 engaging

2.51

 engaged

2.37

 engagement

2.05

 diseng

1.54

 engagements

1.51

 indulge

1.27

 participate

1.25

 partake

1.23

Activations Density 0.022%

verbs related to interaction or conflict

instances of the word "engage" in various contexts

No Known Activations