INDEX

Explanations

words related to weapons and military activities

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

gres

-0.81

 Strawberry

-0.80

dit

-0.73

lus

-0.72

noon

-0.69

cess

-0.68

iannopoulos

-0.67

forth

-0.66

atta

-0.66

Cow

-0.66

POSITIVE LOGITS

powder

1.08

 arsenal

0.94

 weapons

0.93

 wielded

0.91

 caches

0.86

 weaponry

0.85

 weapon

0.85

 manufacturer

0.83

 salute

0.81

 rifles

0.80

Activations Density 0.030%

words related to weapons and military activities

No Comments

No Known Activations