INDEX

Explanations

statements of accusation or blame directed towards someone

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

æĸ¹

-0.86

 migr

-0.72

redo

-0.72

arkable

-0.69

daq

-0.69

ibaba

-0.68

 pilgrimage

-0.67

iosis

-0.67

urgical

-0.66

oubted

-0.64

POSITIVE LOGITS

 unfairly

0.96

 inaction

0.85

 negligence

0.80

 wrongly

0.79

 failing

0.77

 foul

0.76

 faulty

0.76

 dishonest

0.73

 inappropriately

0.72

 corrupt

0.71

Activations Density 0.155%

statements of accusation or blame directed towards someone

No Comments

No Known Activations