huggingface/smollm

View on GitHub

Everything about the SmolLM and SmolVLM family of models

Stars: 3.9k
Forks: 315
Open beginner issues: 0
Indexed issues: 52
Avg merge: 1m
Merged PRs (30d): 1

Dominant language: Python
License: Apache-2.0
Last GitHub push: Apr 2, 2026
Latest indexed: Sep 17, 2026
Contributing guide: No contributing guide
Code of conduct: No code of conduct
Beginner labels: No beginner labels indexed

Open All

52 open issues indexed

Suggestion: We can add huggingface candle inference. Open

Text

huggingface/smollm#23 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Sentencepiece tokenizer Open

Text

huggingface/smollm#29 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 17, 2026
How do you track validation loss, while finetuning the SmolVLM using the finetuning notebook given. Open

Image

huggingface/smollm#30 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
SmolVLM-2B is taking too long to finetune on NVIDIA-A40 around 0.02 iteration/seconds. Open

Image

huggingface/smollm#31 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 17, 2026
One node with 8 GPUs - finetuning notebook error = RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:1 and cuda:0! Open

Image

huggingface/smollm#32 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Request for Smollm1-135M/360M/1.7B's pre-decay nanotron ckpt Open

Text

huggingface/smollm#33 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Request: Evaluation Code for SmolLM-1 Open

Text

huggingface/smollm#34 · 14 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
SmolLM2 Pretrain Dataset Open

Text

huggingface/smollm#35 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
distilabel base pipeline code fails Open

Text

huggingface/smollm#43 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Reproduce pretrained-smollm Open

Text

huggingface/smollm#57 · 1 comment · 1 reaction · 0 assignees · Indexed Sep 17, 2026
SmolVLM as Feature Extractor : seems very promising Open

Image

huggingface/smollm#59 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Converting SmolVLM to ONNX Open

Image

huggingface/smollm#60 · 9 comments · 1 reaction · 0 assignees · Indexed Sep 17, 2026
SmolLM in Spanish Open

Text

huggingface/smollm#63 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Reproducing scripts for fine-tuning SmolLM to SmolVLM Open

Image

huggingface/smollm#64 · 0 comments · 1 reaction · 0 assignees · Indexed Sep 17, 2026
How to set video size when fine tuning Open

Video

huggingface/smollm#65 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
How to fine tune smolvlm on OCR Open

Image

huggingface/smollm#67 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Finetuning SmolVLM2 on multiple tasks Open

Image

huggingface/smollm#69 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Question about the collect_fn in fine-tuning demo Open

Image

huggingface/smollm#70 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
smol-tools Open

Text

huggingface/smollm#71 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Question about Learned Token Open

Image

huggingface/smollm#72 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
datasets of SmolVLM Open

Image

huggingface/smollm#73 · 2 comments · 1 reaction · 0 assignees · Indexed Sep 17, 2026
A float16 tensor's data must be type of function Float16Array() Open

Text

huggingface/smollm#74 · 1 comment · 1 reaction · 0 assignees · Indexed Sep 17, 2026
Unable to replicate any of the SmolLM experiments (due to UnionMatchError when attempting to run training with nanotron library) Open

Text

huggingface/smollm#77 · 4 comments · 1 reaction · 0 assignees · Indexed Sep 17, 2026
how to continously pretrain VLM base model Open

Image Video

huggingface/smollm#78 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
the structure of eval_preds in compute_metrics Open

huggingface/smollm#80 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Tokenizer is lossy - can't tokenize every sequence Open

Text

huggingface/smollm#81 · 1 comment · 1 reaction · 0 assignees · Indexed Sep 17, 2026
smolvlm2 fintue ipynb file is broken Open

Video

huggingface/smollm#82 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Question Regarding SmolVLM2-2.2B-Instruct Fine-tuning Open

Image Video

huggingface/smollm#83 · 0 comments · 1 reaction · 0 assignees · Indexed Sep 17, 2026
Can you provide the file of this environment dependency package? Open

Video

huggingface/smollm#84 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 17, 2026
[Collaboration Proposal] Applying SmolLM principles to "Crane-01," a Sovereign Model for African Language Open

Text

huggingface/smollm#86 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
SmolLM3 distilation plans Open

Text

huggingface/smollm#101 · 2 comments · 1 reaction · 0 assignees · Indexed Sep 17, 2026
deepspeed error: AttributeError: 'LlamaAttention' object has no attribute 'num_heads' Open

Image

huggingface/smollm#103 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Can you give smollm dataset preprocess process? Open

Text

huggingface/smollm#109 · 9 comments · 1 reaction · 0 assignees · Indexed Sep 17, 2026
Access error while reading from s3 bucket Open

Text

huggingface/smollm#113 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 17, 2026
BFCL fork not visible Open

Text

huggingface/smollm#114 · 2 comments · 1 reaction · 0 assignees · Indexed Sep 17, 2026
SmolLM2 SFT training Open

huggingface/smollm#115 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Some datasets in Phase 3 pretraining are repeated Thousands of Times? Open

huggingface/smollm#116 · 0 comments · 1 reaction · 0 assignees · Indexed Sep 17, 2026
Don't support Zero1 Open

huggingface/smollm#117 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Question about SmolLM2 intermediate checkpoints Open

huggingface/smollm#120 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Reproducing the results for SmolVLM Open

huggingface/smollm#121 · 0 comments · 5 reactions · 0 assignees · Indexed Sep 17, 2026
Intermediate checkpoints for SmolVLM Open

huggingface/smollm#122 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
SmolLM3-M2-Interface-Multimodale Open

huggingface/smollm#123 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Why is SmolVLM inference slow Open

huggingface/smollm#125 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
[QUESTION] Can SmolLM3 be used with SmolVLM? Open

huggingface/smollm#126 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Sharing smolLM3-500M base checkpoint (Apache-2.0) Open

huggingface/smollm#127 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
The finetune about bbox Open

huggingface/smollm#128 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Unexpected behavior of models in terms of generation time Open

huggingface/smollm#132 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Score gap when reproducing results across SmolVLM family Open

huggingface/smollm#133 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Significant discrepancy in MATH dataset evaluation: Local results (28.5) vs. Official reported (46.1) for SmolLM3-3B-Base Open

huggingface/smollm#134 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Showcase / question: a board-proven offline language runtime on ESP32-C3, and whether some compact language capability may eventually move beyond small dense models Open

huggingface/smollm#135 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026

huggingface/smollm

Get new issues in your inbox