INDEX

Explanations

phrases related to expressing a clear and direct message

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

phrases that introduce an explanation or a definition.

oai_token-act-pair · gemini-1.5-flash 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

egal

-0.75

 extensively

-0.73

illard

-0.70

undai

-0.67

etheus

-0.66

ibal

-0.64

ockets

-0.63

udeau

-0.63

luster

-0.63

roach

-0.62

POSITIVE LOGITS

tons

0.93

ified

0.85

 mortals

0.83

IFIED

0.77

 coincidence

0.75

ifiable

0.72

 plat

0.71

clair

0.70

wallet

0.70

ton

0.69

Activations Density 0.516%

phrases related to expressing a clear and direct message

phrases that introduce an explanation or a definition.

No Comments

No Known Activations