Search results

Performance decrease after finetuning on 8*A800 GPUs. Got 0.367 in aime24_nofigures task

Status: Open.

#123 In simplescaling/s1;

· LiZhangMing opened on Jun 26, 2025

Performance Drop on MGSM After Finetuning - QWEN2.5 - 3B

Status: Open.

#122 In simplescaling/s1;

· baselmousi opened on Jun 24, 2025

Evaluation of local fine-tuned checkpoint is ~10× slower than the same-weight original Hugging Face model

Status: Open.

#118 In simplescaling/s1;

· SondosBsharat opened on May 10, 2025

budget forcing

Status: Open.

#117 In simplescaling/s1;

· Kenwwww opened on May 9, 2025

1.5B accuracy 0 after SFT

Status: Open.

#115 In simplescaling/s1;

· alvinshao0313 opened on Apr 19, 2025

what is the content of "results/gemini/{subdir}/*.json"

Status: Open.

#114 In simplescaling/s1;

· Pyjacc opened on Apr 18, 2025

Is the code data used in the training data?

Status: Open.

#112 In simplescaling/s1;

· xuhu0115 opened on Apr 8, 2025

Training time shows 18 hr on 16 * H100

Status: Open.

#111 In simplescaling/s1;

· kartikjain-42 opened on Apr 4, 2025

Config for smaller models

Status: Open.

#110 In simplescaling/s1;

· DoubtedSteam opened on Apr 1, 2025

Why generate more than 8k tokens at a time?

Status: Open.

#109 In simplescaling/s1;

· JaydencoolCC opened on Apr 1, 2025

Issue Reproducing s1.1-32B Training Loss (Observed vs. WandB)

Status: Open.

#108 In simplescaling/s1;

· dzh19990407 opened on Mar 31, 2025

Can I specify the thinking context of the query when s1 inferencing with vllm?

Status: Open.

#107 In simplescaling/s1;

· Siki-cloud opened on Mar 24, 2025