INDEX

Explanations

personal identity-related words like "I", "who I was", and "who I am"

oai_token-act-pair · gpt-3.5-turbo

references to self-identity and personal state

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

arya

-0.68

dar

-0.63

 clearance

-0.62

 advance

-0.60

azar

-0.58

 corrobor

-0.58

 breeze

-0.58

 aback

-0.57

wave

-0.57

adv

-0.55

POSITIVE LOGITS

ues

0.73

ufact

0.67

 selves

0.65

 entreprene

0.62

Self

0.62

ternally

0.61

 nutshell

0.61

versions

0.61

 destined

0.59

uably

0.58

Activations Density 0.123%

personal identity-related words like "I", "who I was", and "who I am"

references to self-identity and personal state

No Comments

No Known Activations