INDEX

Explanations

words related to fire and destruction

oai_token-act-pair · gpt-3.5-turbo

references to flames and fire

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

 Zimmer

-0.68

 Nass

-0.66

 PLUS

-0.65

ajor

-0.64

Smart

-0.64

 Visa

-0.62

â

-0.62

Sing

-0.61

imil

-0.61

Som

-0.61

POSITIVE LOGITS

 flames

3.91

 flame

2.83

 Flames

2.13

flame

2.07

 blaze

2.00

 fire

1.85

 fires

1.84

 flaming

1.71

fires

1.69

 ashes

1.66

Activations Density 0.020%

words related to fire and destruction

references to flames and fire

No Comments

No Known Activations