--- license: apache-2.0 base_model: - mistralai/Mistral-Small-3.1-24B-Instruct-2503 base_model_relation: quantized pipeline_tag: text2text-generation language: - zho - eng - fra - spa - por - deu - ita - rus - jpn - kor - vie - tha - ara --- # Elastic model: Mistral-Small-3.1-24B-Instruct-2503. Fastest and most flexible models for self-serving. Elastic models are the models produced by TheStage AI ANNA: Automated Neural Networks Accelerator. ANNA allows you to control model size, latency and quality with a simple slider movement. For each model, ANNA produces a series of optimized models: * __XL__: Mathematically equivalent neural network, optimized with our DNN compiler. * __L__: Near lossless model, with less than 1% degradation obtained on corresponding benchmarks. * __M__: Faster model, with accuracy degradation less than 1.5%. * __S__: The fastest model, with accuracy degradation less than 2%. __Goals of elastic models:__ * Provide flexibility in cost vs quality selection for inference * Provide clear quality and latency benchmarks * Provide interface of HF libraries: transformers and diffusers with a single line of code * Provide models supported on a wide range of hardware, which are pre-compiled and require no JIT. * Provide the best models and service for self-hosting. > It's important to note that specific quality degradation can vary from model to model. For instance, with an S model, you can have 0.5% degradation as well. ![Performance Graph](images/performance_graph.png) ----- ## Inference To infer our models, you just need to replace `transformers` import with `elastic_models.transformers`: ```python import torch from transformers import AutoTokenizer from elastic_models.transformers import AutoModelForCausalLM # Currently we require to have your HF token # as we use original weights for part of layers and # model configuration as well model_name = "mistralai/Mistral-Small-3.1-24B-Instruct-2503" hf_token = '' device = torch.device("cuda") # Create mode tokenizer = AutoTokenizer.from_pretrained( model_name, token=hf_token ) model = AutoModelForCausalLM.from_pretrained( model_name, token=hf_token, torch_dtype=torch.bfloat16, attn_implementation="sdpa", mode='S' ).to(device) model.generation_config.pad_token_id = tokenizer.eos_token_id # Inference simple as transformers library prompt = "Describe basics of DNNs quantization." messages = [ { "role": "system", "content": "You are a search bot, answer on user text queries." }, { "role": "user", "content": prompt } ] chat_prompt = tokenizer.apply_chat_template( messages, add_generation_prompt=True, tokenize=False ) inputs = tokenizer(chat_prompt, return_tensors="pt") inputs.to(device) with torch.inference_mode(): generate_ids = model.generate(**inputs, max_length=500) input_len = inputs['input_ids'].shape[1] generate_ids = generate_ids[:, input_len:] output = tokenizer.batch_decode( generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False )[0] # Validate answer print(f"# Q:\n{prompt}\n") print(f"# A:\n{output}\n") ``` __System requirements:__ * GPUs: H100, L40s * CPU: AMD, Intel * Python: 3.10-3.12 To work with our models just run these lines in your terminal: ```shell pip install thestage pip install elastic_models[nvidia]\ --index-url https://thestage.jfrog.io/artifactory/api/pypi/pypi-thestage-ai-production/simple\ --extra-index-url https://pypi.nvidia.com\ --extra-index-url https://pypi.org/simple pip install flash_attn==2.7.3 --no-build-isolation pip uninstall apex ``` Then go to [app.thestage.ai](https://app.thestage.ai), login and generate API token from your profile page. Set up API token as follows: ```shell thestage config set --api-token ``` Congrats, now you can use accelerated models! ---- ## Benchmarks Benchmarking is one of the most important procedures during model acceleration. We aim to provide clear performance metrics for models using our algorithms. The `W8A8, int8 column` indicates that we applied W8A8 quantization with int8 data type to all linear layers and used the same calibration data as for ANNA. The S model achieves practically identical speed but much higher quality, as ANNA knows how to improve quantization quality on sensitive layers! ### Quality benchmarks | Metric/Model | S | M | L | XL | Original | W8A8, int8 | |---------------|---|---|---|----|----------|------------| | arc_challenge | 65.30 | 66.30 | 66.70 | 66.80 | 66.80 | 51.10 | - | | gsm8k | 87.70 | 88.40 | 87.70 | 88.86 | 88.86 | 13.49 | - | | mmlu | 79.00 | 79.40 | 79.70 | 80.20 | 80.20 | 60.45 | - | | piqa | 82.90 | 83.10 | 82.60 | 83.00 | 83.00 | 75.35 | - | | winogrande | 78.20 | 79.40 | 79.30 | 79.50 | 79.50 | 71.19 | - | * **MMLU**: Evaluates general knowledge across 57 subjects including science, humanities, engineering, and more. Shows model's ability to handle diverse academic topics. * **PIQA**: Evaluates physical commonsense reasoning through questions about everyday physical interactions. Shows model's understanding of real-world physics concepts. * **Arc Challenge**: Evaluates grade-school level multiple-choice questions requiring reasoning. Shows model's ability to solve complex reasoning tasks. * **Winogrande**: Evaluates commonsense reasoning through sentence completion tasks. Shows model's capability to understand context and resolve ambiguity. * **GSM8K**: GSM8K (Grade School Math 8K) is a dataset of 8.5K high quality linguistically diverse grade school math word problems. ### Latency benchmarks ### Performance by Context Size The tables below show performance (tokens per second) for different input context sizes across different GPU models and batch sizes: **H100:** *Batch Size 1:* | Context | Input Tokens | S | M | L | XL | Original | |---------|-------------|---|---|---|----|---------| | Small | 256 | 90.3 | 82.5 | 72.2 | 54.4 | 41.2 | - | | Medium | 1024 | 90.1 | 82.2 | 71.8 | - | 38.8 | - | | Large | 4096 | 88.2 | 81.0 | 70.4 | - | 33.8 | - | *Batch Size 8:* | Context | Input Tokens | S | M | L | XL | Original | |---------|-------------|---|---|---|----|---------| | Small | 256 | 86.5 | 79.9 | 69.1 | - | 36.7 | - | | Medium | 1024 | 80.3 | 74.9 | 65.1 | - | 29.0 | - | | Large | 4096 | 63.3 | 59.5 | 53.1 | - | 15.5 | - | *Batch Size 16:* | Context | Input Tokens | S | M | L | XL | Original | |---------|-------------|---|---|---|----|---------| | Small | 256 | 84.7 | 78.1 | 68.0 | - | 32.2 | - | | Medium | 1024 | 79.8 | 73.3 | 64.1 | - | 21.8 | - | | Large | 4096 | 62.5 | 58.1 | 52.7 | - | 9.7 | - | **L40S:** *Batch Size 1:* | Context | Input Tokens | S | M | L | XL | Original | |---------|-------------|---|---|---|----|---------| | Small | 256 | 26.0 | 24.0 | 21.0 | - | - | - | | Medium | 1024 | 25.8 | 23.8 | 20.9 | - | - | - | | Large | 4096 | 25.1 | 23.3 | 20.5 | - | - | - | *Batch Size 8:* | Context | Input Tokens | S | M | L | XL | Original | |---------|-------------|---|---|---|----|---------| | Small | 256 | 25.2 | 23.2 | 20.4 | - | - | - | | Medium | 1024 | 24.3 | 22.4 | 19.8 | - | - | - | | Large | 4096 | - | - | - | - | - | - | *Batch Size 16:* | Context | Input Tokens | S | M | L | XL | Original | |---------|-------------|---|---|---|----|---------| | Small | 256 | 24.5 | 22.6 | 19.9 | - | - | - | | Medium | 1024 | 22.8 | 20.9 | - | - | - | - | | Large | 4096 | - | - | - | - | - | - | *Note: Results show tokens per second (TPS) for text generation with 100 new tokens output. Performance varies based on GPU model, context size, and batch size.* ## Links * Platform: [app.thestage.ai](https://app.thestage.ai/) * __Subscribe for updates__: [TheStageAI X](https://x.com/TheStageAI) * __Contact email__: contact@thestage.ai