INDEX

Explanations

texts related to question and answer (Q&A) formats

oai_token-act-pair · gpt-3.5-turbo

occurrences of Q&A sessions or related formats in the text

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

ufact

-0.91

esm

-0.73

outhern

-0.71

fell

-0.70

fitted

-0.69

intendent

-0.66

dust

-0.65

fitting

-0.64

lit

-0.63

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.63

POSITIVE LOGITS

 Questions

1.06

 Answers

1.02

 answ

1.00

FAQ

0.98

Answer

0.95

 questions

0.92

answer

0.91

answered

0.90

Discuss

0.85

 session

0.82

Activations Density 0.129%

texts related to question and answer (Q&A) formats

occurrences of Q&A sessions or related formats in the text

No Comments

No Known Activations