Instructions to use Premchan369/Q-TensorFormer with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Premchan369/Q-TensorFormer with Transformers:

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

pipe = pipeline("text-generation", model="Premchan369/Q-TensorFormer", trust_remote_code=True)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("Premchan369/Q-TensorFormer", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Premchan369/Q-TensorFormer with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Premchan369/Q-TensorFormer"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Premchan369/Q-TensorFormer",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Premchan369/Q-TensorFormer

SGLang

How to use Premchan369/Q-TensorFormer 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 "Premchan369/Q-TensorFormer" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Premchan369/Q-TensorFormer",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "Premchan369/Q-TensorFormer" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Premchan369/Q-TensorFormer",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Premchan369/Q-TensorFormer with Docker Model Runner:
```
docker model run hf.co/Premchan369/Q-TensorFormer
```

Q-TensorFormer / FUNCTIONAL_VALIDATION_REPORT.md

Premchandyadav369

fix(config): Align configuration API with canonical ModelConfig fields and document checkpoint provenance

0b7ee79 about 1 month ago

preview code

Raw History Blame Contribute Delete

2.24 kB

# FUNCTIONAL VALIDATION REPORT: Deterministic Runtime Verification

**Date**: September 17, 2026  
**Auditor**: Senior ML Systems Engineer & Research Software Auditor  
**Final Status**: **`FUNCTIONAL_PASS`**
---
## 1. Tokenizer Status
```
TOKENIZER_STATUS = NOT_FOUND
TRACK_A_QUALITY_STATUS = UNAVAILABLE
```
- **Finding**: No pretrained tokenizer vocabulary files (`tokenizer.json`, `vocab.json`, `spm.model`) exist in the repository.
- **Protocol**: Per research software audit requirements, no external tokenizer was fabricated or substituted. All inputs for functional and latency testing are valid integer token tensors within the vocabulary bounds: $x \in \{0, \dots, V-1\}$ where $V = 10,000$.
---
## 2. Checkpoint Loading Gate Status
- **Target**: `qtensorformer_small.pt`
- **Result**: `CHECKPOINT_INCOMPATIBLE` (Proven in `CHECKPOINT_INTEGRITY_REPORT.md`).
- **Policy**: Functional validation is executed using the canonical architectural definition initialized deterministically, avoiding random hyperparameter guessing or silent weight substitution.
---
## 3. Shape & Finite Output Verification
Deterministic forward passes were executed across varying batch and sequence dimensions on CPU:
| Input Batch ($B$) | Sequence ($T$) | Integer Token Range | QTF Output Shape | Dense Output Shape | Output Dtype | Finite Check (NaN/Inf) | Status |
| :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: |
| 1 | 1 | $[0, 10000)$ | `[1, 1, 10000]` | `[1, 1, 10000]` | `torch.float32` | ALL FINITE (0 NaN, 0 Inf) | **PASS** |
| 1 | 8 | $[0, 10000)$ | `[1, 8, 10000]` | `[1, 8, 10000]` | `torch.float32` | ALL FINITE (0 NaN, 0 Inf) | **PASS** |
| 1 | 32 | $[0, 10000)$ | `[1, 32, 10000]` | `[1, 32, 10000]` | `torch.float32` | ALL FINITE (0 NaN, 0 Inf) | **PASS** |

---

## 4. Architectural Invariants Verified
1. **Batch Dimension**: Preserved across all forward layers without reshaping artifacts.
2. **Sequence Dimension**: Output sequence dimension matches input sequence dimension exactly.
3. **Vocabulary Dimension**: Logits projection maps to `vocab_size = 10000`.
4. **Numerical Stability**: Outputs contain zero `NaN` or `Inf` values across all test configurations.

**FINAL STATUS**: **`FUNCTIONAL_PASS`**

	# FUNCTIONAL VALIDATION REPORT: Deterministic Runtime Verification

	Date: September 17, 2026
	Auditor: Senior ML Systems Engineer & Research Software Auditor
	Final Status: `FUNCTIONAL_PASS`

	---

	## 1. Tokenizer Status

	```
	TOKENIZER_STATUS = NOT_FOUND
	TRACK_A_QUALITY_STATUS = UNAVAILABLE
	```
	- Finding: No pretrained tokenizer vocabulary files (`tokenizer.json`, `vocab.json`, `spm.model`) exist in the repository.
	- Protocol: Per research software audit requirements, no external tokenizer was fabricated or substituted. All inputs for functional and latency testing are valid integer token tensors within the vocabulary bounds: $x \in \{0, \dots, V-1\}$ where $V = 10,000$.

	---

	## 2. Checkpoint Loading Gate Status

	- Target: `qtensorformer_small.pt`
	- Result: `CHECKPOINT_INCOMPATIBLE` (Proven in `CHECKPOINT_INTEGRITY_REPORT.md`).
	- Policy: Functional validation is executed using the canonical architectural definition initialized deterministically, avoiding random hyperparameter guessing or silent weight substitution.

	---

	## 3. Shape & Finite Output Verification

	Deterministic forward passes were executed across varying batch and sequence dimensions on CPU:

	\| Input Batch ($B$) \| Sequence ($T$) \| Integer Token Range \| QTF Output Shape \| Dense Output Shape \| Output Dtype \| Finite Check (NaN/Inf) \| Status \|
	\| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \| :---: \|
	\| 1 \| 1 \| $[0, 10000)$ \| `[1, 1, 10000]` \| `[1, 1, 10000]` \| `torch.float32` \| ALL FINITE (0 NaN, 0 Inf) \| PASS \|
	\| 1 \| 8 \| $[0, 10000)$ \| `[1, 8, 10000]` \| `[1, 8, 10000]` \| `torch.float32` \| ALL FINITE (0 NaN, 0 Inf) \| PASS \|
	\| 1 \| 32 \| $[0, 10000)$ \| `[1, 32, 10000]` \| `[1, 32, 10000]` \| `torch.float32` \| ALL FINITE (0 NaN, 0 Inf) \| PASS \|

	---

	## 4. Architectural Invariants Verified
	1. Batch Dimension: Preserved across all forward layers without reshaping artifacts.
	2. Sequence Dimension: Output sequence dimension matches input sequence dimension exactly.
	3. Vocabulary Dimension: Logits projection maps to `vocab_size = 10000`.
	4. Numerical Stability: Outputs contain zero `NaN` or `Inf` values across all test configurations.

	FINAL STATUS: `FUNCTIONAL_PASS`