INDEX

Explanations

dialogue punctuated by exclamations and questions

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ciplinary

-0.72

ancial

-0.72

aukee

-0.71

conservancy

-0.71

otype

-0.70

utterstock

-0.70

¥ŀ

-0.68

oÄŁ

-0.67

isphere

-0.67

 segreg

-0.67

POSITIVE LOGITS

 Suddenly

1.02

 Hearing

0.94

 âĢķ

0.91

 ãĢĮ

0.90

 murm

0.89

 whispered

0.89

 Said

0.88

pause

0.86

 Slowly

0.84

>>\

0.84

Activations Density 0.058%

dialogue punctuated by exclamations and questions

No Comments

No Known Activations