INDEX

Explanations

interviews with individuals

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

math

-0.81

ciplinary

-0.77

mil

-0.74

blue

-0.72

 Goods

-0.71

hover

-0.69

axy

-0.68

cil

-0.64

 Haram

-0.64

sels

-0.64

POSITIVE LOGITS

 interview

1.01

ees

0.96

 interviews

0.94

 conducted

0.90

 transcripts

0.88

 aired

0.87

 booth

0.85

Interview

0.84

 Transcript

0.82

 taped

0.81

Activations Density 0.521%

interviews with individuals

No Comments

No Known Activations