INDEX

Explanations

names of political figures

oai_token-act-pair · gpt-3.5-turbo

names of politicians and legislative 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

 realised

-1.09

 realise

-1.07

 analys

-1.04

 analyse

-0.97

 ageing

-0.94

 civilisation

-0.93

 rumours

-0.91

 favourites

-0.90

 maths

-0.88

 colours

-0.85

POSITIVE LOGITS

Senate

1.11

Republican

1.09

Republicans

1.06

Congress

1.00

 bipartisan

1.00

ospons

0.99

PAC

0.97

GOP

0.96

Obama

0.94

Washington

0.94

Activations Density 0.952%

names of political figures

names of politicians and legislative references

No Comments

No Known Activations