INDEX

Explanations

words related to things that are difficult to track, find, or identify

oai_token-act-pair · gpt-3.5-turbo

terms related to irrefutable evidence or undeniable facts

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

aft

-0.75

bomb

-0.72

Den

-0.69

dr

-0.68

Cyn

-0.65

 Horse

-0.65

ser

-0.64

aman

-0.64

abouts

-0.62

tips

-0.62

POSITIVE LOGITS

utable

3.56

 immutable

1.39

Ã±a

1.38

 Kardashian

1.16

ovable

1.08

readable

1.05

uble

1.05

ractive

1.05

ellar

0.96

 opaque

0.91

Activations Density 0.065%

words related to things that are difficult to track, find, or identify

terms related to irrefutable evidence or undeniable facts

No Comments

No Known Activations