INDEX

Explanations

words related to historical and political references

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

enhagen

-0.85

EStream

-0.81

 flush

-0.77

fman

-0.74

aution

-0.74

 Clicker

-0.69

 destro

-0.66

ãĥ¯

-0.66

 indo

-0.65

 Kers

-0.62

POSITIVE LOGITS

anging

1.33

orses

1.24

undred

1.21

oused

1.20

onest

1.20

ouston

1.15

ollow

1.14

aunted

1.14

oney

1.14

ockey

1.13

Activations Density 0.027%

words related to historical and political references

No Comments

No Known Activations