INDEX

Explanations

numbers and measurements expressed as words

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

 Challenges

-0.67

yles

-0.65

 Rica

-0.62

 Machines

-0.62

 embassies

-0.62

sels

-0.62

 Mysteries

-0.62

 Activities

-0.62

 fame

-0.60

 Muse

-0.59

POSITIVE LOGITS

%-

0.93

0.91

 dimensional

0.83

taboola

0.82

 gallon

0.79

bedroom

0.77

istine

0.77

quel

0.75

acious

0.74

 bedroom

0.73

Activations Density 0.776%

numbers and measurements expressed as words

No Comments

No Known Activations