INDEX

Explanations

phrases related to drawing attention, particularly in the context of politics and activism

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

nown

-0.77

ican

-0.75

keep

-0.71

bom

-0.65

unker

-0.63

 curing

-0.61

terday

-0.59

 Replay

-0.59

onte

-0.59

icum

-0.58

POSITIVE LOGITS

 parallels

1.01

strings

0.90

 drawn

0.89

 attention

0.88

 inspiration

0.87

 conclusions

0.86

cards

0.85

 Cosponsors

0.85

card

0.83

 resemb

0.78

Activations Density 0.457%

phrases related to drawing attention, particularly in the context of politics and activism

No Comments

No Known Activations