INDEX

Explanations

specific items or categories within a larger context

oai_token-act-pair · gpt-3.5-turbo

terms related to classifications and structured entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

umbn

-0.65

shirts

-0.64

 Dragons

-0.61

 Letters

-0.59

 Ladies

-0.59

 Shields

-0.57

amaz

-0.57

iths

-0.54

prem

-0.54

 gratitude

-0.54

POSITIVE LOGITS

 imaginable

1.12

 whatsoever

0.73

 conceivable

0.73

Chip

0.70

 except

0.70

 thereof

0.67

herer

0.67

 nodd

0.65

 individually

0.64

Category

0.61

Activations Density 0.376%

specific items or categories within a larger context

terms related to classifications and structured entities

No Comments

No Known Activations