INDEX

Explanations

time-related phrases, particularly mentioning specific lengths of time such as weeks, months, or years

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

 textual

-0.84

 fixation

-0.68

itialized

-0.63

 cumbers

-0.62

gib

-0.62

erate

-0.61

 hemor

-0.59

 utilization

-0.59

 sync

-0.59

 continuity

-0.59

POSITIVE LOGITS

å¥³

0.81

 afternoon

0.72

ghan

0.70

 Giul

0.67

 marked

0.67

 ç¥ŀ

0.65

å¹

0.65

 evening

0.64

incial

0.64

âĢ²

0.63

Activations Density 0.612%

time-related phrases, particularly mentioning specific lengths of time such as weeks, months, or years

No Comments

No Known Activations