INDEX

Explanations

questions being answered, often related to technical topics

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

wagen

-0.77

fulness

-0.75

ufact

-0.73

coni

-0.72

nings

-0.67

 readable

-0.66

 Lauder

-0.64

 Nadu

-0.64

 Awakens

-0.61

baugh

-0.58

POSITIVE LOGITS

ues

1.15

UE

1.15

ued

1.10

ubes

1.08

addafi

1.03

atari

1.02

WER

1.01

atar

0.99

uran

0.98

iu

0.93

Activations Density 0.498%

questions being answered, often related to technical topics

No Comments

No Known Activations