INDEX

Explanations

personal experiences and emotions expressed in text

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

trap

-0.72

angan

-0.65

ulum

-0.63

ella

-0.62

onto

-0.62

ÃŃa

-0.61

ascal

-0.61

mort

-0.61

inqu

-0.59

 withd

-0.59

POSITIVE LOGITS

 firsthand

1.05

dust

1.03

 VIDEOS

0.88

 parallels

0.88

dogs

0.84

saw

0.82

lights

0.80

 similarities

0.79

tower

0.78

eye

0.77

Activations Density 3.598%

personal experiences and emotions expressed in text

No Comments

No Known Activations