INDEX

Explanations

the word "try" followed by a verb

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

dylib

-0.70

head

-0.66

effect

-0.64

 ---------

-0.61

vation

-0.59

Quality

-0.58

thus

-0.58

ros

-0.58

requisite

-0.58

inational

-0.58

POSITIVE LOGITS

 unsuccessfully

1.27

 harder

0.88

 desperately

0.85

 valiant

0.82

nir

0.78

anke

0.76

 vain

0.74

reprene

0.72

 hard

0.70

 frantically

0.69

Activations Density 1.312%

the word "try" followed by a verb

No Comments

No Known Activations