INDEX

Explanations

references to positions or directions, especially related to the right side or back

oai_token-act-pair · gpt-3.5-turbo

directions and positional references

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ngth

-0.72

upon

-0.70

anmar

-0.69

regor

-0.68

avorite

-0.66

nces

-0.65

phis

-0.64

uctor

-0.63

ritch

-0.63

emort

-0.63

POSITIVE LOGITS

 basis

1.02

 occasions

0.99

 doorstep

0.98

 periphery

0.95

 shelf

0.90

 leash

0.88

 heels

0.88

 occasion

0.88

 behalf

0.83

 pedest

0.82

Activations Density 0.274%

references to positions or directions, especially related to the right side or back

directions and positional references

No Comments

No Known Activations