INDEX

Explanations

phrases related to beer production, historical figures, and hotel management

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

vag

-0.96

amen

-0.94

unc

-0.91

ayed

-0.88

take

-0.84

ighed

-0.84

comed

-0.83

mble

-0.83

olit

-0.82

stretched

-0.81

POSITIVE LOGITS

 Dome

0.81

 Wire

0.81

 Oaks

0.78

 Butt

0.77

 Hyde

0.76

 Cout

0.76

 Tanks

0.76

 Suit

0.75

 Springs

0.75

 Lanka

0.74

Activations Density 1.465%

phrases related to beer production, historical figures, and hotel management

No Comments

No Known Activations