INDEX

Explanations

words related to technical instructions or details, potentially related to technology or programming

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

 Seym

-0.81

hovah

-0.78

enance

-0.72

 Archangel

-0.65

pelling

-0.64

 Seventh

-0.63

 limp

-0.62

 payday

-0.60

 mortg

-0.58

 destro

-0.58

POSITIVE LOGITS

strument

1.60

ned

1.33

ners

1.22

osaurs

1.15

flation

1.12

jury

1.10

omial

1.09

clusive

1.05

ancial

1.05

cluded

1.04

Activations Density 0.764%

words related to technical instructions or details, potentially related to technology or programming

No Comments

No Known Activations