INDEX

Explanations

words related to health and medical conditions

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

 Flavoring

-0.88

mosp

-0.73

ochond

-0.66

rador

-0.60

ortment

-0.60

ools

-0.59

bow

-0.58

icrobial

-0.57

deen

-0.57

arlane

-0.55

POSITIVE LOGITS

 enough

0.99

 nonetheless

0.79

 compared

0.77

0.76

0.74

enough

0.72

:[

0.69

 insofar

0.69

.<

0.68

 nevertheless

0.68

Activations Density 0.224%

words related to health and medical conditions

No Comments

No Known Activations