INDEX

Explanations

names of people or locations

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 unlaw

-0.73

behind

-0.71

sheet

-0.65

anooga

-0.64

 resil

-0.60

modules

-0.60

 disadvantage

-0.60

roach

-0.58

nesday

-0.58

 fundament

-0.56

POSITIVE LOGITS

iewicz

0.80

coni

0.76

oshi

0.72

esis

0.72

ÃŃn

0.70

INAL

0.70

imo

0.69

esi

0.69

inen

0.69

heim

0.67

Activations Density 0.212%

names of people or locations

No Comments

No Known Activations