tarekziade HF Staff
Publish MIC usage, slim fixture and reviewed Transformers expectations
1937d55 verified | # MiniMax-M2.5: causal language modeling with Transformers | |
| This page is the Transformers contract of the MiniMax-M2.5 MIC and also a usage example. [doc-builder](https://github.com/huggingface/doc-builder/blob/main/docs/runnable-code-blocks.md) renders the `runnable` blocks below as documentation, and its pytest plugin runs them as written against `MiniMaxAI/MiniMax-M2.5`. The MIC runner runs the same blocks against each slim fixture and each checkpoint in [`integration.json`](../../integration.json). Lines marked `# doc-builder: hide` only matter there: they swap in the pinned checkpoint or variant and print the result record. | |
| ## Predict the next token | |
| ```py runnable:usage | |
| import torch | |
| from transformers import AutoModelForCausalLM, AutoTokenizer, set_seed | |
| import json # doc-builder: hide | |
| import os # doc-builder: hide | |
| model_id = "MiniMaxAI/MiniMax-M2.5" | |
| model_id = os.environ.get("MIC_MODEL", model_id) # doc-builder: hide | |
| variant = json.loads(os.environ.get("MIC_MODEL_KWARGS", "{}")) # doc-builder: hide | |
| device = "cuda" if torch.cuda.is_available() else "cpu" | |
| set_seed(0) # seeds torch, NumPy, and Python: loading, dropout, and data augmentation draw random numbers | |
| tokenizer = AutoTokenizer.from_pretrained(model_id) | |
| model = AutoModelForCausalLM.from_pretrained(model_id).to(device) | |
| if variant: # doc-builder: hide | |
| del model # doc-builder: hide | |
| model = AutoModelForCausalLM.from_pretrained(model_id, **variant).to(device) # doc-builder: hide | |
| prompts = [ | |
| "The capital of France is", | |
| "The capital of Germany is", | |
| ] | |
| # The tokenizer defines no pad token: pad with its end-of-sequence token, on the left. | |
| tokenizer.pad_token = tokenizer.eos_token | |
| tokenizer.padding_side = "left" | |
| inputs = tokenizer(prompts, return_tensors="pt", padding=True).to(device) | |
| model.eval() | |
| with torch.no_grad(): | |
| logits = model(**inputs).logits | |
| # Probe the last position of each sequence (left padding puts real tokens at the end). | |
| last = inputs["attention_mask"].sum(dim=-1) - 1 | |
| probe = logits[torch.arange(len(prompts)), last][:, :8].float() | |
| print(probe) | |
| ``` | |
| ## One training step | |
| ```py runnable:usage:2 | |
| labels = inputs["input_ids"].clone() | |
| labels[inputs["attention_mask"] == 0] = -100 | |
| model.train() | |
| optimizer = torch.optim.SGD([p for p in model.parameters() if p.requires_grad], lr=0.01) | |
| loss = model(**inputs, labels=labels).loss | |
| loss.backward() | |
| before = {n: p.detach().cpu().clone() for n, p in model.named_parameters() if p.requires_grad} # doc-builder: hide | |
| optimizer.step() | |
| print(f"loss: {loss.item():.4f}") | |
| if os.environ.get("MIC_RUNNER"): # doc-builder: hide | |
| named = dict(model.named_parameters()) | |
| checked = ("model.layers.0.self_attn.q_proj.weight", "model.layers.0.mlp.gate.weight") | |
| result = { | |
| "runtime": { | |
| "device": model.device.type, | |
| "dtype": str(model.dtype).removeprefix("torch."), | |
| "attn_implementation": model.config._attn_implementation, | |
| }, | |
| "logits": {"probe": probe.tolist()}, | |
| "training": { | |
| "loss": loss.item(), | |
| "gradients": {n: bool(named[n].grad is not None and torch.count_nonzero(named[n].grad) > 0 and torch.isfinite(named[n].grad).all()) for n in checked}, | |
| "updated": any(not torch.equal(p.detach().cpu(), before[n]) for n, p in model.named_parameters() if n in before), | |
| }, | |
| } | |
| print(json.dumps(result)) | |
| ``` | |