INDEX

Explanations

terms related to sharing information or content

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

paio

-0.84

 desper

-0.71

ichick

-0.67

annis

-0.64

sil

-0.64

icans

-0.63

²¾

-0.63

 sacked

-0.62

ī

-0.62

wing

-0.61

POSITIVE LOGITS

cro

0.99

ership

0.84

holders

0.84

edin

0.81

holder

0.80

hare

0.79

 similarities

0.77

holding

0.76

ivity

0.70

 shares

0.69

Activations Density 0.721%

terms related to sharing information or content

No Comments

No Known Activations