INDEX

Explanations

historical terms and references

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Hist" or related forms, indicating a focus on historical context or discussions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

PLIED

-0.80

ateral

-0.69

Tube

-0.67

backer

-0.63

 learning

-0.63

zzle

-0.62

[];

-0.62

BOX

-0.60

 schooling

-0.60

 downhill

-0.60

POSITIVE LOGITS

orically

1.39

orical

1.34

oric

1.13

ogram

1.07

alog

1.07

orians

1.06

orie

1.00

orian

0.96

idine

0.94

oria

0.92

Activations Density 0.010%

historical terms and references

instances of the word "Hist" or related forms, indicating a focus on historical context or discussions

No Comments

No Known Activations