INDEX

Explanations

references to heroes

oai_token-act-pair · gpt-3.5-turbo

terms related to the concept of a hero

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ĸļ

-0.87

nap

-0.67

ãģ®å®

-0.66

ocking

-0.65

erer

-0.65

ENTION

-0.64

 discriminating

-0.64

Effective

-0.63

 Edison

-0.63

 racket

-0.60

POSITIVE LOGITS

es

1.00

hero

0.98

ku

0.88

ppo

0.86

complex

0.85

tics

0.81

rities

0.80

ecake

0.78

isodes

0.74

esian

0.72

Activations Density 0.013%

references to heroes

terms related to the concept of a hero

No Comments

No Known Activations