INDEX

Explanations

phrases related to endings or conclusions

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rouse

-0.68

ledged

-0.65

ickr

-0.65

uyomi

-0.65

Recommended

-0.64

iaries

-0.62

iors

-0.62

acted

-0.61

adobe

-0.60

opol

-0.58

POSITIVE LOGITS

 reckoning

0.93

 goodbye

0.89

 hostilities

0.80

 farewell

0.73

 Goodbye

0.71

angering

0.69

THING

0.67

lins

0.66

 humankind

0.65

stem

0.65

Activations Density 0.105%

phrases related to endings or conclusions

No Comments

No Known Activations