INDEX

Explanations

phrases involving actions or activities

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate involvement or engagement in activities or processes

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

Ö¼

-0.70

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.70

 skies

-0.68

ipe

-0.65

onew

-0.64

 ceilings

-0.60

\\\\\\\\\\\\\\\\

-0.60

century

-0.60

 impression

-0.60

 Lamar

-0.60

POSITIVE LOGITS

 involve

0.81

licts

0.77

ioned

0.73

 involving

0.73

itaire

0.73

annel

0.72

 involves

0.72

ulkan

0.71

 involved

0.69

 Deity

0.66

Activations Density 0.020%

phrases involving actions or activities

phrases that indicate involvement or engagement in activities or processes

No Comments

No Known Activations