INDEX

Explanations

phrases related to attempting or making an effort towards achieving something

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

head

-0.74

dylib

-0.67

effect

-0.63

thus

-0.62

ros

-0.61

requisite

-0.60

workers

-0.60

vation

-0.60

resent

-0.60

houses

-0.59

POSITIVE LOGITS

 unsuccessfully

1.36

 harder

0.91

 valiant

0.88

 desperately

0.87

nir

0.79

anke

0.77

 vain

0.76

 frantically

0.75

 dism

0.73

 cram

0.72

Activations Density 3.442%

phrases related to attempting or making an effort towards achieving something

No Comments

No Known Activations