INDEX

Explanations

names of individuals involved in criminal activities or legal cases

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

 slideshow

-0.74

urry

-0.70

 increments

-0.70

lag

-0.70

 Ends

-0.69

ipers

-0.68

 Extend

-0.68

 Which

-0.67

PsyNetMessage

-0.66

inav

-0.66

POSITIVE LOGITS

 married

1.15

 assassinated

1.10

 diagnosed

1.09

 appointed

1.08

 accused

1.07

 stationed

1.07

 credited

1.05

 arrested

1.05

 tasked

1.05

 outspoken

1.04

Activations Density 1.508%

names of individuals involved in criminal activities or legal cases

No Comments

No Known Activations