INDEX

Explanations

mentions of quantities or measurements, specifically in grams

oai_token-act-pair · gpt-3.5-turbo

references to the term "grams."

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

bee

-0.70

 foundation

-0.70

 qualification

-0.69

 failure

-0.66

 mistake

-0.65

aw

-0.65

 origin

-0.64

 pool

-0.62

 licens

-0.62

Yun

-0.62

POSITIVE LOGITS

rams

4.88

ram

1.31

raphic

1.18

raph

1.11

ravings

1.09

rations

1.06

ograms

1.05

raphics

1.01

rals

1.01

iences

1.00

Activations Density 0.012%

mentions of quantities or measurements, specifically in grams

references to the term "grams."

No Comments

No Known Activations