INDEX

Explanations

expressions related to personal reflection, questioning, and opinions

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

oway

-0.74

ilings

-0.71

iling

-0.70

represented

-0.66

avid

-0.65

senal

-0.64

cers

-0.64

 arrivals

-0.63

fter

-0.63

lication

-0.61

POSITIVE LOGITS

 raining

1.34

 happen

0.99

 hurts

0.92

 happened

0.90

 happens

0.81

 happening

0.80

 easier

0.79

 kinda

0.77

 depends

0.74

beh

0.74

Activations Density 1.693%

expressions related to personal reflection, questioning, and opinions

No Comments

No Known Activations