INDEX

Explanations

words related to negative or problematic actions and descriptions

oai_token-act-pair · gpt-3.5-turbo

descriptors and traits associated with action movie characters and their archetypes

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

hower

-0.69

agements

-0.57

 Repeat

-0.57

 petitions

-0.56

 Transcript

-0.54

 Charges

-0.53

ploy

-0.52

usal

-0.51

 absentee

-0.50

RESULTS

-0.50

POSITIVE LOGITS

çĶ°

0.62

ulhu

0.61

Õ

0.59

Ú

0.58

cients

0.57

Û

0.55

ãĥīãĥ©ãĤ´ãĥ³

0.54

Created

0.53

 mascot

0.53

 concept

0.53

Activations Density 2.483%

words related to negative or problematic actions and descriptions

descriptors and traits associated with action movie characters and their archetypes

No Comments

No Known Activations