INDEX

Explanations

phrases related to getting something done or accomplished

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

 Archdemon

-0.66

velt

-0.64

 fallacy

-0.59

 Lies

-0.56

 Gardens

-0.56

 cardinal

-0.55

 Britann

-0.53

cone

-0.53

SPONSORED

-0.53

 Alban

-0.52

POSITIVE LOGITS

rid

1.43

 acquainted

1.14

tin

1.03

cloneembedreportprint

0.97

 married

0.96

 sucked

0.93

 distracted

0.91

bent

0.90

 underway

0.89

 bored

0.88

Activations Density 0.118%

phrases related to getting something done or accomplished

No Comments

No Known Activations