INDEX

Explanations

mentions of or references to physical spaces like rooms

oai_token-act-pair · gpt-3.5-turbo

references to rooms, particularly in a literal or metaphorical context

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

Not in Any Lists

No Comments

Negative Logits

annot

-0.69

 Associ

-0.67

ĻĤ

-0.66

åĪ

-0.63

clipse

-0.63

NF

-0.62

Pwr

-0.61

 Corpor

-0.61

 Governments

-0.60

heid

-0.59

POSITIVE LOGITS

 room

3.85

 Room

2.82

 rooms

2.66

room

2.59

Room

2.39

 Rooms

1.94

rooms

1.88

 hallway

1.53

 chamber

1.48

 bedroom

1.42

Activations Density 0.027%

mentions of or references to physical spaces like rooms

references to rooms, particularly in a literal or metaphorical context

No Comments

No Known Activations