INDEX

Explanations

words related to technology, software, and media

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

ately

-0.74

ALLY

-0.68

ince

-0.65

amental

-0.64

 Citiz

-0.60

ounded

-0.56

 pand

-0.56

ahu

-0.56

osate

-0.56

acco

-0.56

POSITIVE LOGITS

suit

0.96

balls

0.85

suits

0.83

 Tracks

0.82

ball

0.81

tracks

0.75

record

0.75

 tracks

0.74

runner

0.74

bone

0.73

Activations Density 0.974%

words related to technology, software, and media

No Comments

No Known Activations