INDEX

Explanations

specific locations or names of places

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

hower

-0.79

ļéĨĴ

-0.77

 "$:/

-0.74

ical

-0.69

vernment

-0.68

ICAL

-0.65

NCT

-0.61

ITED

-0.60

ãģĦ

-0.58

hound

-0.58

POSITIVE LOGITS

rict

1.16

onew

1.07

roller

1.06

alker

1.03

omach

0.98

uart

0.97

amped

0.96

amping

0.96

okes

0.96

oppable

0.95

Activations Density 2.920%

specific locations or names of places

No Comments

No Known Activations