INDEX

Explanations

references to returning or recovering something in various contexts

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

ussen

-0.79

Cola

-0.65

iop

-0.63

 disse

-0.63

ghai

-0.62

bis

-0.62

chen

-0.62

esson

-0.61

stood

-0.59

asm

-0.58

POSITIVE LOGITS

ees

1.00

 home

0.96

home

0.87

 HOME

0.87

 postage

0.74

 safely

0.72

ported

0.70

prise

0.69

 Home

0.69

owship

0.68

Activations Density 0.301%

references to returning or recovering something in various contexts

No Comments

No Known Activations