INDEX

Explanations

dates formatted as month followed by year

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

ãĤ´ãĥ³

-0.83

ãĤ¦ãĤ¹

-0.75

alo

-0.72

arily

-0.72

arah

-0.69

SourceFile

-0.67

Otherwise

-0.65

ãĥ¥

-0.63

Widget

-0.61

Done

-0.60

POSITIVE LOGITS

 however

1.08

 meanwhile

0.80

when

0.78

 when

0.77

 moreover

0.73

 though

0.68

 according

0.67

 researchers

0.66

 tensions

0.65

we

0.64

Activations Density 0.734%

dates formatted as month followed by year

No Comments

No Known Activations