INDEX

Explanations

phrases indicating novelty or surprise

oai_token-act-pair · gpt-3.5-turbo

phrases indicating experiences of novelty or uniqueness

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

Tips

-0.70

lear

-0.67

reg

-0.65

ŃĶ

-0.64

fund

-0.63

supp

-0.62

haus

-0.61

absor

-0.61

ERT

-0.61

relations

-0.60

POSITIVE LOGITS

 anything

0.95

 anybody

0.88

 anyone

0.81

 daylight

0.77

ANY

0.71

 anywhere

0.70

 dime

0.70

 them

0.68

nor

0.68

him

0.67

Activations Density 0.070%

phrases indicating novelty or surprise

phrases indicating experiences of novelty or uniqueness

No Comments

No Known Activations