INDEX

Explanations

mentions of specific names, places, and organizations

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

 possible

-0.65

 adoptive

-0.60

 distance

-0.58

 invention

-0.58

 sunshine

-0.57

 responsibility

-0.54

 independ

-0.54

 multim

-0.54

 icing

-0.53

 TODAY

-0.52

POSITIVE LOGITS

pillar

0.88

spir

0.86

uthor

0.82

asta

0.81

Ã±o

0.81

hammer

0.79

EStream

0.78

alle

0.77

Corp

0.76

kamp

0.76

Activations Density 1.642%

mentions of specific names, places, and organizations

No Comments

No Known Activations