INDEX

Explanations

detailing of military and conflict-related events and incidents

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

ãĤ´ãĥ³

-0.97

drawn

-0.74

reci

-0.72

bet

-0.70

Craft

-0.70

soType

-0.70

cano

-0.69

venture

-0.68

chart

-0.67

zl

-0.66

POSITIVE LOGITS

 unsuspecting

1.24

 innocent

1.22

 civilians

1.17

 unarmed

1.00

 targets

0.92

 innoc

0.91

him

0.84

 indiscrim

0.83

 anyone

0.82

 havoc

0.82

Activations Density 2.741%

detailing of military and conflict-related events and incidents

No Comments

No Known Activations