INDEX

Explanations

instructions related to practical tasks or hobbies

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

andals

-0.79

bernatorial

-0.79

amaz

-0.77

missing

-0.77

blems

-0.75

ptoms

-0.71

akings

-0.70

 woes

-0.70

comings

-0.69

ocious

-0.68

POSITIVE LOGITS

 confines

0.88

 kios

0.83

 rather

0.80

medium

0.80

 medium

0.78

 predetermined

0.77

 pasture

0.76

 preset

0.76

 proprietary

0.74

 canvas

0.73

Activations Density 3.503%

instructions related to practical tasks or hobbies

No Comments

No Known Activations