Audio-Text-to-Text
Transformers
Safetensors
edgeinstant
feature-extraction
audio
text-to-speech
custom_code
Instructions to use chenjz24/EdgeIn-v3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use chenjz24/EdgeIn-v3 with Transformers:
# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("chenjz24/EdgeIn-v3", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
Download training_config.json from chenjz24/EdgeIn-v3: direct link, hf CLI and curl.
- Browser
- Download file 2.34 kB
-
https://huggingface.co/chenjz24/EdgeIn-v3/resolve/main/training_config.json
- Command line
-
hf download hf://chenjz24/EdgeIn-v3/training_config.json
-
curl -L -o training_config.json https://huggingface.co/chenjz24/EdgeIn-v3/resolve/main/training_config.json
2.34 kB
| { | |
| "model": "/default-filesys/workspace/qintian/runs/S1_fullopen_aut24_s05v6_lv6bx3_40_20260918/final", | |
| "manifests": [ | |
| "/default-filesys/workspace/chenjunzhe/EdgeInstant-1.5b/data/bilingual_s2_s10_v3_clean/train.jsonl" | |
| ], | |
| "output": "/default-filesys/workspace/chenjunzhe/EdgeInstant-1.5b/runs/s1_5_asr_tothemoon_sft_v3_clean_0920_v2", | |
| "epochs": 1, | |
| "max_steps": 0, | |
| "batch_size": 32, | |
| "max_tokens_per_batch": 8192, | |
| "mixture_epoch_counts": { | |
| "asr_read_request": 209825, | |
| "asr_dialogue": 60950, | |
| "asr_web": 91562, | |
| "asr_diverse": 40634, | |
| "asr_hard": 20317, | |
| "speech_translation": 159231, | |
| "instruction_qa": 324948, | |
| "thinking": 116096, | |
| "sound_understanding": 102802, | |
| "text_instruction": 163237, | |
| "text_translation": 118418, | |
| "repetition_negative": 2940 | |
| }, | |
| "gradient_accumulation_steps": 1, | |
| "learning_rate": 1e-06, | |
| "thinker_learning_rate": 2e-06, | |
| "audio_tower_learning_rate": 1e-07, | |
| "loss_normalization": "example", | |
| "eos_loss_weight": 2.0, | |
| "unlikelihood_weight": 0.2, | |
| "lr_schedule": "cosine", | |
| "warmup_ratio": 0.03, | |
| "weight_decay": 0.01, | |
| "max_grad_norm": 1.0, | |
| "max_length": 4096, | |
| "precision": "bf16", | |
| "train_thinker": "full", | |
| "audio_request_lora": false, | |
| "last_n_layers": 4, | |
| "lora_r": 16, | |
| "lora_alpha": 32, | |
| "lora_dropout": 0.0, | |
| "lora_target_modules": [ | |
| "q_proj", | |
| "k_proj", | |
| "v_proj", | |
| "o_proj", | |
| "in_proj_qkv", | |
| "in_proj_z", | |
| "out_proj", | |
| "gate_proj", | |
| "up_proj", | |
| "down_proj" | |
| ], | |
| "train_audio_tower": false, | |
| "train_audio_tower_last_n": 2, | |
| "train_audio_projector": true, | |
| "gradient_checkpointing": false, | |
| "num_workers": 4, | |
| "group_by_length": true, | |
| "seed": 260914210, | |
| "save_every": 800, | |
| "eval_config": null, | |
| "eval_every": null, | |
| "eval_at_start": false, | |
| "validation_manifests": [ | |
| "/default-filesys/workspace/chenjunzhe/EdgeInstant-1.5b/data/bilingual_s2_s10_v3_clean/validation.jsonl" | |
| ], | |
| "validation_every": 400, | |
| "early_stopping_patience": 0, | |
| "early_stopping_min_delta": 0.001, | |
| "wandb": true, | |
| "wandb_project": "audio-in", | |
| "wandb_entity": "chenjz24-tsinghua-university", | |
| "wandb_group": "s1_5_asr_tothemoon_sft_v3_clean_0920_v2", | |
| "run_name": "s1_5_asr_tothemoon_sft_v3_clean_0920_v2", | |
| "log_every": 10, | |
| "distillation_teacher": null, | |
| "distillation_weight": 0.3 | |
| } | |