INDEX

Explanations

Roman numerals indicating levels of importance or hierarchy

oai_token-act-pair · gpt-3.5-turbo

references to a person or entity named "Il."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

enance

-0.64

ELL

-0.61

AA

-0.57

 altogether

-0.57

ARK

-0.57

edIn

-0.55

tons

-0.54

 quickly

-0.53

eele

-0.53

POSITIVE LOGITS

ibrary

1.13

uminati

1.05

legal

1.05

ustration

1.03

ove

1.00

BuyableInstoreAndOnline

0.94

usive

0.93

leg

0.85

umin

0.85

iberal

0.82

Activations Density 0.040%

Roman numerals indicating levels of importance or hierarchy

references to a person or entity named "Il."

No Comments

No Known Activations

Roman numerals indicating levels of importance or hierarchy

references to a person or entity named "Il."

No Comments

No Known Activations