INDEX

Explanations

phrases related to high-ranking positions in organizations or government

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

ier

-0.66

 breeze

-0.65

Joined

-0.64

 strain

-0.63

 mosquit

-0.61

 tumble

-0.58

rouse

-0.58

 crumble

-0.58

acci

-0.58

 treated

-0.57

POSITIVE LOGITS

 communications

0.86

 Operations

0.85

mbuds

0.80

 Communications

0.79

ĳå£«

0.77

²¾

0.76

 Offic

0.74

 Strategic

0.74

¥µ

0.74

 Languages

0.71

Activations Density 0.554%

phrases related to high-ranking positions in organizations or government

No Comments

No Known Activations