INDEX

Explanations

phrases indicating a specific intention or purpose

oai_token-act-pair · gpt-3.5-turbo

references to intent 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 @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Tycoon

-0.72

 Chocobo

-0.68

 Cinderella

-0.67

 Ibid

-0.64

Occ

-0.63

cub

-0.62

stood

-0.62

 Cats

-0.62

 Jenner

-0.62

 Serge

-0.61

POSITIVE LOGITS

 intent

1.29

intent

0.88

 intention

0.85

 intending

0.84

lessly

0.83

illery

0.82

 guiActiveUn

0.79

fulness

0.79

edly

0.75

 Intent

0.75

Activations Density 0.005%

phrases indicating a specific intention or purpose

references to intent in various contexts

No Comments

No Known Activations