INDEX

Explanations

mentions of durations or time-related information

oai_token-act-pair · gpt-3.5-turbo

concepts related to time duration and intervals

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.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

ĪĴ

-0.77

 pestic

-0.65

 kindred

-0.65

ropolitan

-0.64

 Gunn

-0.62

 Volks

-0.61

 Cosponsors

-0.61

âĵĺ

-0.60

 sovere

-0.59

STATE

-0.59

POSITIVE LOGITS

 shorter

1.41

 leng

1.34

 shortest

1.33

 elapsed

1.30

 longest

1.29

 duration

1.28

 longer

1.28

 milliseconds

1.26

duration

1.23

seconds

1.22

Activations Density 0.411%

mentions of durations or time-related information

concepts related to time duration and intervals

No Comments

No Known Activations