INDEX

Explanations

technical descriptions or details related to physical objects

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate observation or noticing details

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 farewell

-0.64

 goodbye

-0.63

Gim

-0.63

gress

-0.62

 pseudonym

-0.62

aiden

-0.62

icipated

-0.60

 handling

-0.60

 Serving

-0.59

 ACTION

-0.59

POSITIVE LOGITS

 realize

0.97

 realise

0.96

 discern

0.95

 noticed

0.90

uggest

0.83

 discover

0.82

 reveals

0.81

 noticing

0.81

 discovers

0.80

 realized

0.80

Activations Density 0.300%

technical descriptions or details related to physical objects

phrases that indicate observation or noticing details

No Comments

No Known Activations