INDEX

Explanations

words related to geometric shapes and medical conditions (like 'tetrahedron')

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

é¾įå

-0.71

BILITIES

-0.66

DEF

-0.60

 Wolves

-0.58

 STOR

-0.56

ework

-0.56

 Bucks

-0.56

CTR

-0.55

asted

-0.55

 Somalia

-0.55

POSITIVE LOGITS

tons

0.99

apeake

0.89

ding

0.85

sis

0.81

esville

0.81

witz

0.80

nard

0.80

ochond

0.80

nut

0.80

ĪĴ

0.79

Activations Density 0.036%

words related to geometric shapes and medical conditions (like 'tetrahedron')

No Comments

No Known Activations