INDEX

Explanations

hand-drawn maps or diagrams within text

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

ndra

-0.94

jury

-0.78

ership

-0.77

uay

-0.73

weeney

-0.73

nuts

-0.73

--+

-0.71

nder

-0.69

aneous

-0.68

onte

-0.68

POSITIVE LOGITS

ographer

0.93

map

0.88

 coordinates

0.85

 mapping

0.81

 visualization

0.80

 maps

0.79

 mapped

0.78

 diagram

0.77

 tiles

0.74

Map

0.72

Activations Density 0.989%

hand-drawn maps or diagrams within text

No Comments

No Known Activations