INDEX

Explanations

guests or guests' names in a talk show or interview setting

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

cling

-0.80

croft

-0.73

ŃĶ

-0.71

baugh

-0.70

 spect

-0.68

rings

-0.67

rums

-0.65

hower

-0.65

pillar

-0.63

 womb

-0.63

POSITIVE LOGITS

vernment

1.24

errilla

1.12

inea

1.11

itars

1.08

bernatorial

1.07

essing

1.03

arant

1.02

cci

1.00

zman

0.98

ilt

0.98

Activations Density 0.558%

guests or guests' names in a talk show or interview setting

No Comments

No Known Activations