INDEX

Explanations

names starting with "qu" or "qua"

oai_token-act-pair · gpt-3.5-turbo

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

croft

-0.68

strap

-0.64

nod

-0.62

 dividing

-0.62

AGES

-0.62

 testosterone

-0.61

 warming

-0.59

 contraception

-0.58

fulness

-0.58

 tolerate

-0.57

POSITIVE LOGITS

isition

1.16

ipment

1.03

ois

1.02

irrel

1.01

ilibrium

0.99

isite

0.98

isitions

0.98

isites

0.97

arie

0.96

ests

0.96

Activations Density 0.020%

names starting with "qu" or "qua"

No Known Activations