Instructions to use DJLougen/Qwen3.5-9B-Math-GraphCalc with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use DJLougen/Qwen3.5-9B-Math-GraphCalc with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="DJLougen/Qwen3.5-9B-Math-GraphCalc")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("DJLougen/Qwen3.5-9B-Math-GraphCalc")
model = AutoModelForCausalLM.from_pretrained("DJLougen/Qwen3.5-9B-Math-GraphCalc", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use DJLougen/Qwen3.5-9B-Math-GraphCalc with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "DJLougen/Qwen3.5-9B-Math-GraphCalc"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "DJLougen/Qwen3.5-9B-Math-GraphCalc",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/DJLougen/Qwen3.5-9B-Math-GraphCalc

SGLang

How to use DJLougen/Qwen3.5-9B-Math-GraphCalc with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "DJLougen/Qwen3.5-9B-Math-GraphCalc" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "DJLougen/Qwen3.5-9B-Math-GraphCalc",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "DJLougen/Qwen3.5-9B-Math-GraphCalc" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "DJLougen/Qwen3.5-9B-Math-GraphCalc",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use DJLougen/Qwen3.5-9B-Math-GraphCalc with Docker Model Runner:
```
docker model run hf.co/DJLougen/Qwen3.5-9B-Math-GraphCalc
```

DJLougen commited on about 23 hours ago

Commit

6b554ee

verified ·

1 Parent(s): 9b8659b

Add AIME 2025 & AIME 2026 official evaluations + 7B-14B peer benchmark comparison

Browse files

Files changed (1) hide show

README.md +75 -11

README.md CHANGED Viewed

@@ -101,10 +101,41 @@ model-index:
           name: HuggingFaceH4/aime_2024
           type: HuggingFaceH4/aime_2024
           split: train
         metrics:
           - type: accuracy
             value: 10.00
             name: ChatML Zero-Shot Accuracy (%)
 ---
 <p align="center">
@@ -281,22 +312,26 @@ Across the 60 unseen hard-math and function-graphing problems, `Qwen3.5-9B-Math-
 | `Qwen3.5-9B-Math-Tutor` (`32L WSP`) | `2,916,073` | `0` | `81.67%` | `75.00%` | `53.92%` | `0.00%` | `0.00%` |
 | **`Qwen3.5-9B-Math-GraphCalc` (`43L`)** | **`2,920,169`** | **`1,045,984`** | **`96.67%`** | **`70.00%`** | **`97.06%`** | **`100.00%`** | **`100.00%`** |
-### 2. Official Public HuggingFace Math Benchmarks (`7,546` Total Zero-Shot Greedy Evaluations)
 <p align="center">
-  <img src="assets/hf_public_benchmarks.png" alt="Official Public HuggingFace Math Benchmarks — GSM8K, MATH-500, AIME 2024, and MATH-500 Computational-50" width="100%">
 </p>
-Evaluated zero-shot with greedy decoding directly on official HuggingFace Hub datasets ([`openai/gsm8k`](https://huggingface.co/datasets/openai/gsm8k) full `1,319` test set, [`HuggingFaceH4/MATH-500`](https://huggingface.co/datasets/HuggingFaceH4/MATH-500) full `500` competition set, and [`HuggingFaceH4/aime_2024`](https://huggingface.co/datasets/HuggingFaceH4/aime_2024) full `30` Olympiad set across both raw `Plain Q&A` and `ChatML` formats):
 | Public HF Dataset | Arm | Mean Math Acc | Plain Q&A Acc | ChatML Acc | Plain Verified Teach | ChatML Verified Teach |
 |---|---|---:|---:|---:|---:|---:|
-| **`openai/gsm8k`** (`N = 1,319 × 2`) | `Qwen3.5-9B-Base` | `78.54%` | `82.79%` | `74.30%` | `51.40%` | `5.31%` |
-| **`openai/gsm8k`** (`N = 1,319 × 2`) | **`9B WSP Tutor / GraphCalc`** | **`75.78%`** | **`72.71%`** | **`78.85%`** | **`59.89%`** | **`53.07%`** |
-| **`HuggingFaceH4/MATH-500`** (`N = 500 × 2`) | `Qwen3.5-9B-Base` | `48.40%` | `40.60%` | `56.20%` | `17.60%` | `2.80%` |
-| **`HuggingFaceH4/MATH-500`** (`N = 500 × 2`) | **`9B WSP Tutor / GraphCalc`** | **`51.80%`** | **`45.40%`** | **`58.20%`** | **`28.00%`** | **`42.40%`** |
-| **`HuggingFaceH4/aime_2024`** (`N = 30 × 2`) | `Qwen3.5-9B-Base` | `6.67%` | `3.33%` | `10.00%` | `0.00%` | `0.00%` |
-| **`HuggingFaceH4/aime_2024`** (`N = 30 × 2`) | **`9B WSP Tutor / GraphCalc`** | **`8.34%`** | **`6.67%`** | **`10.00%`** | **`6.67%`** | **`0.00%`** |
 #### `MATH-500` Accuracy by Competition Difficulty Level (Zero-Shot ChatML)
@@ -308,9 +343,38 @@ Evaluated zero-shot with greedy decoding directly on official HuggingFace Hub da
 | Level 4 | `48.44%` | **`52.34%`** |
 | Level 5 | `29.10%` | **`29.85%`** |
-> **Domain Scope (`<tools>` vs. Standard Zero-Shot):** `graphcalc.elf` implements a specialized single-variable graphing & 256-bit arithmetic/CORDIC calculator (`+ - * / ^`, `sin`, `cos`, `tan`, `sqrt`, `ln`, `exp`, roots, extrema, Simpson integrals). Pass `<tools>` for multi-digit arithmetic, exact fractions, calculus, and function graphing (**`96.67%`** accuracy on `GraphCalc-HeldOut-60`). For general symbolic competition math, modular number theory, or multi-variable systems (where forcing a single-variable `<tool_call>` on a 50-item `MATH-500` scalar slice scored `58.00%` vs `72.00%` without `<tools>` across `2,927,967` RV32 steps), omit `<tools>` — the untied `E6` gate triggers **`0 / 360`** false tool calls and runs at full 32-layer speed.
-### 3. Disjoint Zero-Shot Math Tutoring & Language Retention (`N = 180 × 2 = 360` Generations, No `<tools>`)
 When prompted without `<tools>`, untied `lm_head` `<tool_call>` calibration (`E6 = 1.3×`) triggers **`0 / 360` false tool calls** while improving zero-shot `plain` Q&A tutoring and preserving `FineWeb-Edu` causal language modeling (`+0.0001` nats):

           name: HuggingFaceH4/aime_2024
           type: HuggingFaceH4/aime_2024
           split: train
+        metrics:
+          - type: accuracy
+            value: 20.00
+            name: ChatML Zero-Shot Accuracy (%)
+          - type: accuracy
+            value: 16.67
+            name: Mean Accuracy Plain+ChatML (%)
+      - task:
+          type: text-generation
+          name: AIME 2025 Olympiad Benchmark (Zero-Shot Greedy)
+        dataset:
+          name: MathArena/aime_2025
+          type: MathArena/aime_2025
+          split: train
+        metrics:
+          - type: accuracy
+            value: 20.00
+            name: ChatML Zero-Shot Accuracy (%)
+          - type: accuracy
+            value: 13.33
+            name: Mean Accuracy Plain+ChatML (%)
+      - task:
+          type: text-generation
+          name: AIME 2026 Olympiad Benchmark (Zero-Shot Greedy)
+        dataset:
+          name: MathArena/aime_2026
+          type: MathArena/aime_2026
+          split: train
         metrics:
           - type: accuracy
             value: 10.00
             name: ChatML Zero-Shot Accuracy (%)
+          - type: accuracy
+            value: 11.67
+            name: Mean Accuracy Plain+ChatML (%)
 ---
 <p align="center">
 | `Qwen3.5-9B-Math-Tutor` (`32L WSP`) | `2,916,073` | `0` | `81.67%` | `75.00%` | `53.92%` | `0.00%` | `0.00%` |
 | **`Qwen3.5-9B-Math-GraphCalc` (`43L`)** | **`2,920,169`** | **`1,045,984`** | **`96.67%`** | **`70.00%`** | **`97.06%`** | **`100.00%`** | **`100.00%`** |
+### 2. Official Public HuggingFace Math Benchmarks (`7,786` Total Zero-Shot Greedy Evaluations)
 <p align="center">
+  <img src="assets/hf_public_benchmarks.png" alt="Official Public HuggingFace Math Benchmarks — GSM8K, MATH-500, AIME 2024, AIME 2025, and AIME 2026" width="100%">
 </p>
+Evaluated zero-shot with greedy decoding directly on official HuggingFace Hub datasets ([`openai/gsm8k`](https://huggingface.co/datasets/openai/gsm8k) full `1,319` test set, [`HuggingFaceH4/MATH-500`](https://huggingface.co/datasets/HuggingFaceH4/MATH-500) full `500` competition set, [`HuggingFaceH4/aime_2024`](https://huggingface.co/datasets/HuggingFaceH4/aime_2024), [`MathArena/aime_2025`](https://huggingface.co/datasets/MathArena/aime_2025), and [`MathArena/aime_2026`](https://huggingface.co/datasets/MathArena/aime_2026) Olympiad sets across both raw `Plain Q&A` and `ChatML` formats):
 | Public HF Dataset | Arm | Mean Math Acc | Plain Q&A Acc | ChatML Acc | Plain Verified Teach | ChatML Verified Teach |
 |---|---|---:|---:|---:|---:|---:|
+| **[`openai/gsm8k`](https://huggingface.co/datasets/openai/gsm8k)** (`N = 1,319 × 2`) | `Qwen3.5-9B-Base` | `78.54%` | `82.79%` | `74.30%` | `51.40%` | `5.31%` |
+| **[`openai/gsm8k`](https://huggingface.co/datasets/openai/gsm8k)** (`N = 1,319 × 2`) | **`9B WSP Tutor / GraphCalc`** | **`75.78%`** | **`72.71%`** | **`78.85%`** | **`59.89%`** | **`53.07%`** |
+| **[`HuggingFaceH4/MATH-500`](https://huggingface.co/datasets/HuggingFaceH4/MATH-500)** (`N = 500 × 2`) | `Qwen3.5-9B-Base` | `48.40%` | `40.60%` | `56.20%` | `17.60%` | `2.80%` |
+| **[`HuggingFaceH4/MATH-500`](https://huggingface.co/datasets/HuggingFaceH4/MATH-500)** (`N = 500 × 2`) | **`9B WSP Tutor / GraphCalc`** | **`51.80%`** | **`45.40%`** | **`58.20%`** | **`28.00%`** | **`42.40%`** |
+| **[`HuggingFaceH4/aime_2024`](https://huggingface.co/datasets/HuggingFaceH4/aime_2024)** (`N = 30 × 2`) | `Qwen3.5-9B-Base` | `13.33%` | `6.67%` | `20.00%` | `3.33%` | `0.00%` |
+| **[`HuggingFaceH4/aime_2024`](https://huggingface.co/datasets/HuggingFaceH4/aime_2024)** (`N = 30 × 2`) | **`9B WSP Tutor / GraphCalc`** | **`16.67%`** | **`13.33%`** | **`20.00%`** | **`13.33%`** | **`3.33%`** |
+| **[`MathArena/aime_2025`](https://huggingface.co/datasets/MathArena/aime_2025)** (`N = 30 × 2`) | `Qwen3.5-9B-Base` | `8.33%` | `0.00%` | `16.67%` | `0.00%` | `3.33%` |
+| **[`MathArena/aime_2025`](https://huggingface.co/datasets/MathArena/aime_2025)** (`N = 30 × 2`) | **`9B WSP Tutor / GraphCalc`** | **`13.33%`** | **`6.67%`** | **`20.00%`** | **`6.67%`** | **`13.33%`** |
+| **[`MathArena/aime_2026`](https://huggingface.co/datasets/MathArena/aime_2026)** (`N = 30 × 2`) | `Qwen3.5-9B-Base` | `6.67%` | `3.33%` | `10.00%` | `3.33%` | `0.00%` |
+| **[`MathArena/aime_2026`](https://huggingface.co/datasets/MathArena/aime_2026)** (`N = 30 × 2`) | **`9B WSP Tutor / GraphCalc`** | **`11.67%`** | **`13.33%`** | **`10.00%`** | **`13.33%`** | **`3.33%`** |
 #### `MATH-500` Accuracy by Competition Difficulty Level (Zero-Shot ChatML)
 | Level 4 | `48.44%` | **`52.34%`** |
 | Level 5 | `29.10%` | **`29.85%`** |
+> **Domain Scope (`<tools>` vs. Standard Zero-Shot):** `graphcalc.elf` implements a specialized single-variable graphing & 256-bit arithmetic/CORDIC calculator (`+ - * / ^`, `sin`, `cos`, `tan`, `sqrt`, `ln`, `exp`, roots, extrema, Simpson integrals). Pass `<tools>` for multi-digit arithmetic, exact fractions, calculus, and function graphing (**`96.67%`** accuracy on `GraphCalc-HeldOut-60`). For general symbolic competition math, modular number theory, or multi-variable systems (where forcing a single-variable `<tool_call>` on a 50-item `MATH-500` scalar slice scored `58.00%` vs `72.00%` without `<tools>` across `2,927,967` RV32 steps), omit `<tools>` — the untied `E6` gate triggers **`0`** false tool calls across all zero-shot evaluations and runs at full 32-layer speed.
+### 3. Peer Model Comparison on Public HF & Technical-Report Benchmarks
+<p align="center">
+  <img src="assets/model_comparison.png" alt="Peer Comparison: Qwen3.5-9B-Math-GraphCalc (Zero-Gradient WSP) vs. 7B-14B Pretrained Base and SFT+RL Models" width="100%">
+</p>
+Comparing **`Qwen3.5-9B-Math-GraphCalc`** (`0` gradient steps, `0` SFT, `0` RL, evaluated **0-shot greedy**) against published `7B–14B` pretrained base peers (`0` post-training gradients) and gradient post-trained (`SFT + RL`) reference models from official technical reports ([`arXiv:2505.09388`](https://arxiv.org/abs/2505.09388), [`arXiv:2412.15115`](https://arxiv.org/abs/2412.15115), [`arXiv:2409.12122`](https://arxiv.org/abs/2409.12122), [`arXiv:2503.19786`](https://arxiv.org/abs/2503.19786)) and [`MathArena`](https://huggingface.co/MathArena):
+| Model | Post-Training | Eval Protocol | GSM8K | MATH / MATH-500 | AIME 2024 | AIME 2025 | AIME 2026 |
+|---|---|---|---:|---:|---:|---:|---:|
+| `Mistral-7B-v0.1` | **0-Grad Base** | 8s / 4s | `36.2%` | `10.2%` | — | — | — |
+| `Meta-Llama-3.1-8B` | **0-Grad Base** | 8s / 4s | `57.2%` | `20.3%` | — | — | — |
+| `DeepSeekMath-Base-7B` | **0-Grad Base** | 8s / 4s | `64.2%` | `36.2%` | — | — | — |
+| `Gemma-2-9B` | **0-Grad Base** | 5s / 4s | `68.6%` | `36.6%` | — | — | — |
+| `Gemma-3-12B-PT` (`12B`) | **0-Grad Base** | 8s / 4s | `71.0%` | `43.3%` | — | — | — |
+| `Qwen2.5-7B` | **0-Grad Base** | 8s / 4s | `85.36%` | `49.80%` | — | — | — |
+| `Qwen2.5-Math-7B` | **0-Grad Base** | 8s / 4s | `91.6%` | `55.4%` | — | — | — |
+| `Qwen2.5-14B` (`14B`) | **0-Grad Base** | 8s / 4s | `90.22%` | `55.64%` | — | — | — |
+| `Qwen3-8B-Base` | **0-Grad Base** | 8s / 4s | `89.84%` | `60.80%` | — | — | — |
+| `Qwen/Qwen3.5-9B-Base` | **0-Grad Base** | **0-shot ChatML** | `74.30%` | `56.20%` | `20.00%` | `16.67%` | `10.00%` |
+| **`Qwen3.5-9B-Math-GraphCalc` (Ours)** | **0-Grad WSP + RV32** | **0-shot ChatML** | **`78.85%`** | **`58.20%`** | **`20.00%`** | **`20.00%`** | **`10.00%`** |
+| `Llama-3.1-8B-Instruct` | SFT + RL | Non-Think (`≤2K`) | `84.5%` | `54.8%` | `6.3%` | `2.7%` | — |
+| `Qwen2.5-7B-Instruct` | SFT + RL | Non-Think (`≤2K`) | `91.6%` | `77.6%` | `9.1%` | `12.1%` | — |
+| `Qwen2.5-Math-7B-Instruct` | SFT + GRPO RL | CoT (`≤2K`) | `95.2%` | `83.6%` | `16.7%` | — | — |
+| `Gemma-3-12B-IT` (`12B`) | SFT + RL Distill | Non-Think (`≤2K`) | — | `85.6%` | `22.4%` | `18.8%` | — |
+| `Qwen3-8B` (Non-Thinking) | SFT + 4-Stage RL | Non-Think (`≤2K`) | — | `87.4%` | `29.1%` | `20.9%` | — |
+| `Qwen3-8B` (Thinking) | SFT + 4-Stage RL | `32K` `<think>` | — | `97.4%` | `76.0%` | `67.3%` | — |
+| `Qwen3.5-9B` (Thinking) | SFT + Scaled RL | `31K` avg `<think>` | — | — | — | — | `92.50%` |
+### 4. Disjoint Zero-Shot Math Tutoring & Language Retention (`N = 180 × 2 = 360` Generations, No `<tools>`)
 When prompted without `<tools>`, untied `lm_head` `<tool_call>` calibration (`E6 = 1.3×`) triggers **`0 / 360` false tool calls** while improving zero-shot `plain` Q&A tutoring and preserving `FineWeb-Edu` causal language modeling (`+0.0001` nats):