INDEX

Explanations

instances where something is being done or needs to be done

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

ipel

-0.70

aten

-0.66

 Arri

-0.65

 Torn

-0.64

 Sect

-0.63

ioxide

-0.61

illi

-0.61

 corridors

-0.60

 sshd

-0.60

 passages

-0.60

POSITIVE LOGITS

 wrong

0.94

 differently

0.94

 else

0.92

 proactive

0.89

 rash

0.88

wrong

0.85

else

0.81

 naughty

0.81

 drastic

0.80

 unethical

0.80

Activations Density 0.052%

instances where something is being done or needs to be done

No Comments

No Known Activations