INDEX

Explanations

the word "stick" following by another word

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

ãĤ¡

-0.79

dinand

-0.76

 unfocusedRange

-0.72

zed

-0.68

terday

-0.66

une

-0.66

ordan

-0.64

ãĥĨãĤ£

-0.63

æ©Ł

-0.63

Planet

-0.63

POSITIVE LOGITS

rily

0.90

holder

0.84

 plaster

0.80

holders

0.80

iness

0.77

pole

0.76

 blender

0.75

ciating

0.74

tails

0.74

raft

0.73

Activations Density 0.757%

the word "stick" following by another word

No Comments

No Known Activations