INDEX

Explanations

mentions of the word "smoker", particularly in relation to various contexts

oai_token-act-pair · gpt-3.5-turbo

references to poker

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

syn

-0.67

 Grain

-0.64

 tasting

-0.63

 nurt

-0.63

 sauce

-0.62

ynt

-0.62

recy

-0.60

 Inspect

-0.60

 USDA

-0.58

 contracting

-0.58

POSITIVE LOGITS

oker

4.47

okers

2.72

oke

1.10

okes

1.02

oked

1.01

 Joker

0.98

izza

0.92

ucker

0.85

ocker

0.84

roud

0.84

Activations Density 0.027%

mentions of the word "smoker", particularly in relation to various contexts

references to poker

No Comments

No Known Activations