INDEX

Explanations

legal and courtroom-related terms

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

 Juno

-0.65

 Carbuncle

-0.64

psy

-0.64

cler

-0.62

beam

-0.60

ources

-0.59

cy

-0.59

ecycle

-0.58

hook

-0.58

POSITIVE LOGITS

 against

1.25

 convinc

1.05

 vehemently

1.04

against

1.03

 forcefully

1.02

 persu

0.96

Against

0.91

 arguing

0.88

 loudly

0.87

 Against

0.86

Activations Density 0.601%

legal and courtroom-related terms

No Comments

No Known Activations