Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models. Built upon extensive training, Qwen3 delivers groundbreaking advancements in reasoning, instruction-following, agent capabilities, and multilingual support, with the following key features: - Uniquely support of seamless switching between thinking mode (for complex logical reasoning, math, and coding) and non-thinking mode (for efficient, general-purpose dialogue) within single model, ensuring optimal performance across various scenarios. - Significantly enhancement in its reasoning capabilities, surpassing previous QwQ (in thinking mode) and…
Open-weight model · Text generation
qwen3-8b-sft-261001-bigmath-sol-fsdp-0-epoch-4
by Seungyeop Yi devpotatopotato/qwen3-8b-sft-261001-bigmath-sol-fsdp-0-epoch-4
qwen3-8b-sft-261001-bigmath-sol-fsdp-0-epoch-4 is an open-weight model for text generation from Seungyeop Yi, released under Apache License 2.0. It has 4.1B parameters and a 40,960-token context. At 16-bit it needs about 9.8 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index.
Qwen3-8B after full-parameter supervised fine-tuning for mathematical keyword and meaning generation. This is the exact saved epoch 4 checkpoint at optimizer step 2976.
Parameters4.1B
Context40,960
Weights32.8 GB
Licenseapache-2.0
AccessOpen weights
Monthly Downloads—
Runs On
What it takes to serve qwen3-8b-sft-261001-bigmath-sol-fsdp-0-epoch-4 (4.1B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 8.2 GB | 9.8 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 4.1 GB | 4.9 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 2.0 GB | 2.5 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Oct 7, 2026.
Model Card
By Seungyeop Yi, published under apache-2.0, revision a41a11bcf12c.
Qwen3-8B after full-parameter supervised fine-tuning for mathematical keyword and meaning generation. This is the exact saved epoch 4 checkpoint at optimizer step 2976. The repository contains the model weights, configuration, and tokenizer files for inference. Training dataset: devpotatopotato/math-keyword-training, keyword-261001-bigmath-gpt-6-sol.jsonl. Use the tokenizer chat template with enablethinking=False. Pass the following template as a user message and replace {problem} with the problem text
Read Seungyeop Yi's full model card
Qwen3-8B after full-parameter supervised fine-tuning for mathematical keyword and meaning generation.
This is the exact saved epoch 4 checkpoint at optimizer step 2976. The repository contains the model weights, configuration, and tokenizer files for inference.
Training dataset: devpotatopotato/math-keyword-training, keyword-261001-bigmath-gpt-6-sol.jsonl.
Use the tokenizer chat template with enable_thinking=False. Pass the following template as a user message and replace {problem} with the problem text:
Read the problem and identify the most important idea for solving it. Express the idea as a keyword or short phrase, and explain its meaning in detail.
Requirements:
- Return the single most useful keyword and its corresponding meaning.
- Choose concrete, specific insights, methods, reductions, constructions, or theorems that guide the best solution path. Avoid vague or broad terms.
- The keyword must contain one to five short words. Prefer fewer words.
- Make the meaning as detailed and comprehensive as possible. Define the keyword, describe its relevant properties, and clarify why it is useful. The meaning must stand on its own, without referring to the specific problem or including any solution steps, attempts, or problem-specific applications.
- Do not include a separate final answer, a reasoning trace, Markdown, or any text outside the required tags.
- Follow the output format:
"""
<keyword>keyword</keyword>
<meaning>meaning</meaning>
"""
Problem:
{problem}
Configuration
- Architecture
- Qwen3ForCausalLM
- Context length (tokens)
- 40,960
- Layers
- 36
- Hidden size
- 4,096
- Feed-forward size
- 12,288
- Attention heads
- 32
- Key/value heads
- 8
- Head dimension
- 128
- Vocabulary size
- 151,936
- RoPE base
- 1,000,000
- Model type
- qwen3
Identity and Version
- Repository
- devpotatopotato/qwen3-8b-sft-261001-bigmath-sol-fsdp-0-epoch-4
- Publisher
- Seungyeop Yi
- Task
- Text generation
- Modality
- Text
- Library
- transformers
- Parameters
- 4.1B parameters
- Languages
- sft
- Revision
- a41a11bcf12c9d65cdb4aee589864dd74c667b37
- First published
- 2026-10-05
- Last updated
- 2026-10-05
Files and Weights
19 files, 32.8 GB in total. The weights are 7 files totalling 32.8 GB in safetensors.
Weights7 files · 32.8 GB
Configuration5 files · 36.0 KB
Tokenizer4 files · 15.9 MB
Documentation1 file · 1.8 KB
Other1 file · 4.2 KB
Repository1 file · 1.6 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model-00001-of-00007.safetensors | Weights | 5.0 GB | 9244ad3d35ca |
| model-00002-of-00007.safetensors | Weights | 4.8 GB | 6c8b3e51e80b |
| model-00003-of-00007.safetensors | Weights | 4.8 GB | a1315bcb4c2c |
| model-00004-of-00007.safetensors | Weights | 5.0 GB | 8c0d08398dee |
| model-00005-of-00007.safetensors | Weights | 4.8 GB | bed583e2ceac |
| model-00006-of-00007.safetensors | Weights | 4.8 GB | be2d6d66d8f6 |
| model-00007-of-00007.safetensors | Weights | 3.5 GB | 36aef47ec869 |
| added_tokens.json | Configuration | 707 B | — |
| config.json | Configuration | 1.5 KB | — |
| generation_config.json | Configuration | 188 B | — |
| model.safetensors.index.json | Configuration | 32.9 KB | — |
| special_tokens_map.json | Configuration | 613 B | — |
| README.md | Documentation | 1.8 KB | — |
| chat_template.jinja | Other | 4.2 KB | — |
| .gitattributes | Repository | 1.6 KB | — |
| merges.txt | Tokenizer | 1.7 MB | — |
| tokenizer.json | Tokenizer | 11.4 MB | aeb13307a71a |
| tokenizer_config.json | Tokenizer | 5.4 KB | — |
| vocab.json | Tokenizer | 2.8 MB | — |
License and Download
- License
- apache-2.0
- Access
- Open weights, no gate
- Download size
- 32.8 GB
Download from Seungyeop Yi
Released by Seungyeop Yi through its official repository on Hugging Face. Read the license.
Built From
- Derived from Qwen/Qwen3-8B
- Trained on (disclosed) devpotatopotato/math-keyword-training
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 32.8 GB |
| 16-bit | 8.2 GB |
| 8-bit | 4.1 GB |
| 4-bit | 2.0 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About qwen3-8b-sft-261001-bigmath-sol-fsdp-0-epoch-4
How much GPU memory does qwen3-8b-sft-261001-bigmath-sol-fsdp-0-epoch-4 need?
About 9.8 GB at 16-bit and 2.5 GB at 4-bit: the weights (4.1B parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run qwen3-8b-sft-261001-bigmath-sol-fsdp-0-epoch-4 on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use qwen3-8b-sft-261001-bigmath-sol-fsdp-0-epoch-4 commercially?
Yes. qwen3-8b-sft-261001-bigmath-sol-fsdp-0-epoch-4 is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.
What is qwen3-8b-sft-261001-bigmath-sol-fsdp-0-epoch-4's context length?
40,960 tokens, from the maximum position embeddings in its published configuration.
Similar Models
We introduce the updated version of the Qwen3-4B non-thinking mode, named Qwen3-4B-Instruct-2507, featuring the following key enhancements: - Significant improvements in general capabilities, including instruction following, logical reasoning, text comprehension, mathematics, science, coding and tool usage. - Substantial gains in long-tail knowledge coverage across multiple languages. - Markedly better alignment with user preferences in subjective and open-ended tasks, enabling more helpful responses and higher-quality text generation. - Enhanced capabilities in 256K long-context understanding. Qwen3-4B-Instruct-2507 has the following features: NOTE: This model supports only non-thinking…
Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models. Building upon extensive advancements in training data, model architecture, and optimization techniques, Qwen3 delivers the following key improvements over the previously released Qwen2.5: Qwen3-4B-Base has the following features: For more details, including benchmark evaluation, hardware requirements, and inference performance, please refer to our blog, GitHub, and Documentation. The code of Qwen3 has been in the latest Hugging Face transformers and we advise you to use the latest version of transformers. With transformers<4.51.0, you will…
Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models. Building upon extensive advancements in training data, model architecture, and optimization techniques, Qwen3 delivers the following key improvements over the previously released Qwen2.5: Qwen3-4B-Base has the following features: For more details, including benchmark evaluation, hardware requirements, and inference performance, please refer to our blog, GitHub, and Documentation. The code of Qwen3 has been in the latest Hugging Face transformers and we advise you to use the latest version of transformers. With transformers<4.51.0, you will…
SRA-RiskGate-4B is an autonomous risk scoring, compliance verification, and dispute adjudication model fine-tuned on top of Qwen/Qwen3-4B-Instruct-2507. It is engineered for both ends of a stablecoin payment's operational lifecycle: Autonomous on-chain agents can hallucinate payment transfers, sign malformed calldata, or trigger catastrophic transactions during market depegs. The deterministic pre-filter and policy firewall is available directly as a verified Action Provider for the Coinbase AgentKit framework: Register RiskGateActionProvider as the payment provider on your AgentKit instance. It checks chain support, policies, and peg deviations, executing safe ERC-20 transfers only after…
![[HYU_NLP] EVA Team logo](/media/mh-logo-hyu-nlp-eval-558e075b88.webp){kind=logo}
Intermediate policy from dynamic OnlineRubrics-Every GRPO training. Distinct from static-rubric GRPO. Base model: Qwen/Qwen3-4B-Instruct-2507; thinking disabled. This checkpoint is a policy state used by the Phase-1 audit. No downstream medical capability or safety claim is made. Research use only; not validated for clinical decision-making. Root files are the veRL-exported Hugging Face inference model (BF16). originalcheckpoint/ preserves the exact original FSDP parameter checkpoint and tokenizer/configuration files. Optimizer state, training data, responses, rubrics, infrastructure configuration, and credentials are not included. The original is retained because export…