Instructions to use ibm-granite/granite-speech-4.1-2b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use ibm-granite/granite-speech-4.1-2b with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="ibm-granite/granite-speech-4.1-2b")# Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("ibm-granite/granite-speech-4.1-2b") model = AutoModelForMultimodalLM.from_pretrained("ibm-granite/granite-speech-4.1-2b", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- AMD Developer Cloud
| { | |
| "architectures": [ | |
| "GraniteSpeechForConditionalGeneration" | |
| ], | |
| "audio_token_index": 100352, | |
| "downsample_rate": 5, | |
| "dtype": "bfloat16", | |
| "encoder_config": { | |
| "context_size": 200, | |
| "conv_expansion_factor": 2, | |
| "conv_kernel_size": 15, | |
| "dim_head": 128, | |
| "dropout": 0.1, | |
| "dtype": "bfloat16", | |
| "feedforward_mult": 4, | |
| "hidden_dim": 1024, | |
| "input_dim": 160, | |
| "max_pos_emb": 512, | |
| "model_type": "granite_speech_encoder", | |
| "num_heads": 8, | |
| "num_layers": 16, | |
| "output_dim": 348 | |
| }, | |
| "has_lora_adapter": false, | |
| "initializer_range": 0.02, | |
| "model_type": "granite_speech", | |
| "projector_config": { | |
| "_attn_implementation_autoset": true, | |
| "attention_probs_dropout_prob": 0.1, | |
| "cross_attention_frequency": 1, | |
| "dtype": "bfloat16", | |
| "encoder_hidden_size": 1024, | |
| "hidden_act": "gelu", | |
| "hidden_dropout_prob": 0.1, | |
| "hidden_size": 1024, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 4096, | |
| "layer_norm_eps": 1e-12, | |
| "max_position_embeddings": 2048, | |
| "model_type": "blip_2_qformer", | |
| "num_attention_heads": 16, | |
| "num_hidden_layers": 2, | |
| "position_embedding_type": "absolute", | |
| "use_qformer_text_input": false, | |
| "vocab_size": 30522 | |
| }, | |
| "text_config": { | |
| "_name_or_path": "ibm-granite/granite-4.0-1b-base", | |
| "architectures": [ | |
| "GraniteForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "attention_multiplier": 0.0078125, | |
| "bos_token_id": 100257, | |
| "dtype": "bfloat16", | |
| "embedding_multiplier": 12.0, | |
| "eos_token_id": 100257, | |
| "hidden_act": "silu", | |
| "hidden_size": 2048, | |
| "initializer_range": 0.1, | |
| "intermediate_size": 4096, | |
| "logits_scaling": 8.0, | |
| "max_position_embeddings": 4096, | |
| "mlp_bias": false, | |
| "model_type": "granite", | |
| "num_attention_heads": 16, | |
| "num_hidden_layers": 40, | |
| "num_key_value_heads": 4, | |
| "pad_token_id": 100256, | |
| "residual_multiplier": 0.22, | |
| "rms_norm_eps": 1e-05, | |
| "rope_parameters": { | |
| "rope_theta": 10000, | |
| "rope_type": "default" | |
| }, | |
| "rope_scaling": null, | |
| "rope_theta": 10000.0, | |
| "use_cache": true, | |
| "vocab_size": 100353 | |
| }, | |
| "tie_word_embeddings": false, | |
| "transformers_version": "4.57.6", | |
| "window_size": 15 | |
| } | |