INDEX

Explanations

phrases related to specific timelines or sequences

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

uckland

-0.77

ught

-0.77

erv

-0.72

adel

-0.71

mingham

-0.70

alty

-0.70

comed

-0.70

stro

-0.69

shr

-0.67

ersive

-0.66

POSITIVE LOGITS

VII

1.34

 VIII

1.31

III

1.26

IV

1.23

XII

1.22

XIV

1.17

XVI

1.16

IX

1.16

Activations Density 1.074%

phrases related to specific timelines or sequences

No Comments

No Known Activations