INDEX

Explanations

phrases related to sharing information or spreading knowledge

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

glers

-0.86

paio

-0.82

ournal

-0.78

ichick

-0.74

tarians

-0.67

usalem

-0.67

 fres

-0.65

 submar

-0.62

teness

-0.61

berus

-0.61

POSITIVE LOGITS

sheets

0.97

holder

0.86

hare

0.84

ership

0.82

soType

0.79

holders

0.78

edin

0.74

 shares

0.74

flower

0.73

 share

0.73

Activations Density 5.126%

phrases related to sharing information or spreading knowledge

No Comments

No Known Activations