INDEX

Explanations

sequences of numbers mixed with some specific proper nouns and abbreviations

oai_token-act-pair · gpt-3.5-turbo

numerical data and statistics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.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

 automation

-0.63

 newsp

-0.61

 roundup

-0.60

 guiActiveUn

-0.60

 facilit

-0.60

 vide

-0.60

 wheelchair

-0.59

 impro

-0.59

 multip

-0.59

hon

-0.58

POSITIVE LOGITS

 TOTAL

0.89

 respectively

0.86

Tot

0.77

 Actual

0.74

"},"

0.72

 attRot

0.71

 Conclusion

0.71

"}

0.71

 Deaths

0.70

 Total

0.69

Activations Density 0.144%

sequences of numbers mixed with some specific proper nouns and abbreviations

numerical data and statistics

No Comments

No Known Activations