INDEX

Explanations

technical or descriptive terms related to objects, processes, or events

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

agree

-0.84

Ĥİ

-0.79

ften

-0.78

redits

-0.78

mbuds

-0.78

osponsors

-0.78

©¶æ

-0.77

hers

-0.77

uther

-0.77

OIL

-0.77

POSITIVE LOGITS

 nature

0.99

 sounding

0.90

 conco

0.82

 task

0.82

 array

0.82

 sibling

0.81

ly

0.79

 wooden

0.79

 piece

0.79

 Victorian

0.77

Activations Density 1.223%

technical or descriptive terms related to objects, processes, or events

No Comments

No Known Activations