INDEX

Explanations

phrases related to physical actions and instructions

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

 presided

-0.71

eria

-0.68

catentry

-0.64

etimes

-0.63

ylon

-0.61

 span

-0.60

cember

-0.59

 capped

-0.56

 teased

-0.56

 Span

-0.55

POSITIVE LOGITS

 ASAP

0.95

 urgently

0.81

dress

0.78

 THEIR

0.75

ipt

0.74

 their

0.74

heed

0.72

 theirs

0.71

igate

0.70

 Tsukuyomi

0.69

Activations Density 0.315%

phrases related to physical actions and instructions

No Comments

No Known Activations