update

README.md

@@ -1,3 +1,11 @@
 ---
 license: mit
 ---
+
+# LongVA-7B-Qwen2-VoiceAssistant
+
+
+LongVA-7B-Qwen2-VoiceAssistant is an extension of [LongVA-7B](https://github.com/EvolvingLMMs-Lab/LongVA), further trained using the [VoiceAssistant](https://huggingface.co/datasets/gpt-omni/VoiceAssistant-400K) dataset.
+
+
+*Please refer to [M4](https://github.com/patrick-tssn/M4) for more details*.

added_tokens.json

@@ -0,0 +1,5 @@
+{
+  "<|endoftext|>": 151643,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644
+}

config.json

@@ -0,0 +1,861 @@
+{
+  "_name_or_path": "checkpoints/LongVA-7B-Qwen2",
+  "architectures": [
+    "LlavaQwenForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "hidden_act": "silu",
+  "hidden_size": 3584,
+  "image_aspect_ratio": "anyres",
+  "image_crop_resolution": null,
+  "image_grid_pinpoints": [
+    [
+      336,
+      672
+    ],
+    [
+      336,
+      1008
+    ],
+    [
+      336,
+      1344
+    ],
+    [
+      336,
+      1680
+    ],
+    [
+      336,
+      2016
+    ],
+    [
+      336,
+      2352
+    ],
+    [
+      336,
+      2688
+    ],
+    [
+      336,
+      3024
+    ],
+    [
+      336,
+      3360
+    ],
+    [
+      336,
+      3696
+    ],
+    [
+      336,
+      4032
+    ],
+    [
+      336,
+      4368
+    ],
+    [
+      336,
+      4704
+    ],
+    [
+      336,
+      5040
+    ],
+    [
+      336,
+      5376
+    ],
+    [
+      336,
+      5712
+    ],
+    [
+      336,
+      6048
+    ],
+    [
+      336,
+      6384
+    ],
+    [
+      336,
+      6720
+    ],
+    [
+      336,
+      7056
+    ],
+    [
+      336,
+      7392
+    ],
+    [
+      336,
+      7728
+    ],
+    [
+      336,
+      8064
+    ],
+    [
+      336,
+      8400
+    ],
+    [
+      336,
+      8736
+    ],
+    [
+      336,
+      9072
+    ],
+    [
+      336,
+      9408
+    ],
+    [
+      336,
+      9744
+    ],
+    [
+      336,
+      10080
+    ],
+    [
+      336,
+      10416
+    ],
+    [
+      336,
+      10752
+    ],
+    [
+      336,
+      11088
+    ],
+    [
+      336,
+      11424
+    ],
+    [
+      336,
+      11760
+    ],
+    [
+      336,
+      12096
+    ],
+    [
+      336,
+      12432
+    ],
+    [
+      336,
+      12768
+    ],
+    [
+      336,
+      13104
+    ],
+    [
+      336,
+      13440
+    ],
+    [
+      336,
+      13776
+    ],
+    [
+      336,
+      14112
+    ],
+    [
+      336,
+      14448
+    ],
+    [
+      336,
+      14784
+    ],
+    [
+      336,
+      15120
+    ],
+    [
+      336,
+      15456
+    ],
+    [
+      336,
+      15792
+    ],
+    [
+      336,
+      16128
+    ],
+    [
+      336,
+      16464
+    ],
+    [
+      672,
+      336
+    ],
+    [
+      672,
+      672
+    ],
+    [
+      672,
+      1008
+    ],
+    [
+      672,
+      1344
+    ],
+    [
+      672,
+      1680
+    ],
+    [
+      672,
+      2016
+    ],
+    [
+      672,
+      2352
+    ],
+    [
+      672,
+      2688
+    ],
+    [
+      672,
+      3024
+    ],
+    [
+      672,
+      3360
+    ],
+    [
+      672,
+      3696
+    ],
+    [
+      672,
+      4032
+    ],
+    [
+      672,
+      4368
+    ],
+    [
+      672,
+      4704
+    ],
+    [
+      672,
+      5040
+    ],
+    [
+      672,
+      5376
+    ],
+    [
+      672,
+      5712
+    ],
+    [
+      672,
+      6048
+    ],
+    [
+      672,
+      6384
+    ],
+    [
+      672,
+      6720
+    ],
+    [
+      672,
+      7056
+    ],
+    [
+      672,
+      7392
+    ],
+    [
+      672,
+      7728
+    ],
+    [
+      672,
+      8064
+    ],
+    [
+      1008,
+      336
+    ],
+    [
+      1008,
+      672
+    ],
+    [
+      1008,
+      1008
+    ],
+    [
+      1008,
+      1344
+    ],
+    [
+      1008,
+      1680
+    ],
+    [
+      1008,
+      2016
+    ],
+    [
+      1008,
+      2352
+    ],
+    [
+      1008,
+      2688
+    ],
+    [
+      1008,
+      3024
+    ],
+    [
+      1008,
+      3360
+    ],
+    [
+      1008,
+      3696
+    ],
+    [
+      1008,
+      4032
+    ],
+    [
+      1008,
+      4368
+    ],
+    [
+      1008,
+      4704
+    ],
+    [
+      1008,
+      5040
+    ],
+    [
+      1008,
+      5376
+    ],
+    [
+      1344,
+      336
+    ],
+    [
+      1344,
+      672
+    ],
+    [
+      1344,
+      1008
+    ],
+    [
+      1344,
+      1344
+    ],
+    [
+      1344,
+      1680
+    ],
+    [
+      1344,
+      2016
+    ],
+    [
+      1344,
+      2352
+    ],
+    [
+      1344,
+      2688
+    ],
+    [
+      1344,
+      3024
+    ],
+    [
+      1344,
+      3360
+    ],
+    [
+      1344,
+      3696
+    ],
+    [
+      1344,
+      4032
+    ],
+    [
+      1680,
+      336
+    ],
+    [
+      1680,
+      672
+    ],
+    [
+      1680,
+      1008
+    ],
+    [
+      1680,
+      1344
+    ],
+    [
+      1680,
+      1680
+    ],
+    [
+      1680,
+      2016
+    ],
+    [
+      1680,
+      2352
+    ],
+    [
+      1680,
+      2688
+    ],
+    [
+      1680,
+      3024
+    ],
+    [
+      2016,
+      336
+    ],
+    [
+      2016,
+      672
+    ],
+    [
+      2016,
+      1008
+    ],
+    [
+      2016,
+      1344
+    ],
+    [
+      2016,
+      1680
+    ],
+    [
+      2016,
+      2016
+    ],
+    [
+      2016,
+      2352
+    ],
+    [
+      2016,
+      2688
+    ],
+    [
+      2352,
+      336
+    ],
+    [
+      2352,
+      672
+    ],
+    [
+      2352,
+      1008
+    ],
+    [
+      2352,
+      1344
+    ],
+    [
+      2352,
+      1680
+    ],
+    [
+      2352,
+      2016
+    ],
+    [
+      2352,
+      2352
+    ],
+    [
+      2688,
+      336
+    ],
+    [
+      2688,
+      672
+    ],
+    [
+      2688,
+      1008
+    ],
+    [
+      2688,
+      1344
+    ],
+    [
+      2688,
+      1680
+    ],
+    [
+      2688,
+      2016
+    ],
+    [
+      3024,
+      336
+    ],
+    [
+      3024,
+      672
+    ],
+    [
+      3024,
+      1008
+    ],
+    [
+      3024,
+      1344
+    ],
+    [
+      3024,
+      1680
+    ],
+    [
+      3360,
+      336
+    ],
+    [
+      3360,
+      672
+    ],
+    [
+      3360,
+      1008
+    ],
+    [
+      3360,
+      1344
+    ],
+    [
+      3696,
+      336
+    ],
+    [
+      3696,
+      672
+    ],
+    [
+      3696,
+      1008
+    ],
+    [
+      3696,
+      1344
+    ],
+    [
+      4032,
+      336
+    ],
+    [
+      4032,
+      672
+    ],
+    [
+      4032,
+      1008
+    ],
+    [
+      4032,
+      1344
+    ],
+    [
+      4368,
+      336
+    ],
+    [
+      4368,
+      672
+    ],
+    [
+      4368,
+      1008
+    ],
+    [
+      4704,
+      336
+    ],
+    [
+      4704,
+      672
+    ],
+    [
+      4704,
+      1008
+    ],
+    [
+      5040,
+      336
+    ],
+    [
+      5040,
+      672
+    ],
+    [
+      5040,
+      1008
+    ],
+    [
+      5376,
+      336
+    ],
+    [
+      5376,
+      672
+    ],
+    [
+      5376,
+      1008
+    ],
+    [
+      5712,
+      336
+    ],
+    [
+      5712,
+      672
+    ],
+    [
+      6048,
+      336
+    ],
+    [
+      6048,
+      672
+    ],
+    [
+      6384,
+      336
+    ],
+    [
+      6384,
+      672
+    ],
+    [
+      6720,
+      336
+    ],
+    [
+      6720,
+      672
+    ],
+    [
+      7056,
+      336
+    ],
+    [
+      7056,
+      672
+    ],
+    [
+      7392,
+      336
+    ],
+    [
+      7392,
+      672
+    ],
+    [
+      7728,
+      336
+    ],
+    [
+      7728,
+      672
+    ],
+    [
+      8064,
+      336
+    ],
+    [
+      8064,
+      672
+    ],
+    [
+      8400,
+      336
+    ],
+    [
+      8736,
+      336
+    ],
+    [
+      9072,
+      336
+    ],
+    [
+      9408,
+      336
+    ],
+    [
+      9744,
+      336
+    ],
+    [
+      10080,
+      336
+    ],
+    [
+      10416,
+      336
+    ],
+    [
+      10752,
+      336
+    ],
+    [
+      11088,
+      336
+    ],
+    [
+      11424,
+      336
+    ],
+    [
+      11760,
+      336
+    ],
+    [
+      12096,
+      336
+    ],
+    [
+      12432,
+      336
+    ],
+    [
+      12768,
+      336
+    ],
+    [
+      13104,
+      336
+    ],
+    [
+      13440,
+      336
+    ],
+    [
+      13776,
+      336
+    ],
+    [
+      14112,
+      336
+    ],
+    [
+      14448,
+      336
+    ],
+    [
+      14784,
+      336
+    ],
+    [
+      15120,
+      336
+    ],
+    [
+      15456,
+      336
+    ],
+    [
+      15792,
+      336
+    ],
+    [
+      16128,
+      336
+    ],
+    [
+      16464,
+      336
+    ]
+  ],
+  "image_split_resolution": null,
+  "initializer_range": 0.02,
+  "intermediate_size": 18944,
+  "max_position_embeddings": 224000,
+  "max_window_layers": 28,
+  "mm_hidden_size": 1024,
+  "mm_patch_merge_type": "unires",
+  "mm_projector_lr": null,
+  "mm_projector_type": "mlp2x_gelu",
+  "mm_resampler_type": null,
+  "mm_spatial_pool_mode": "average",
+  "mm_spatial_pool_stride": 2,
+  "mm_tunable_parts": "speech_projector,mm_language_model",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "mm_vision_select_feature": "patch",
+  "mm_vision_select_layer": -2,
+  "mm_vision_tower": "checkpoints/clip-vit-large-patch14-336",
+  "mm_vision_tower_lr": 2e-06,
+  "model_type": "qwen2",
+  "num_attention_heads": 28,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 4,
+  "pos_skipping_range": 4096,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000000.0,
+  "sliding_window": null,
+  "speech_encoder": "checkpoints/whisper/large-v3.pt",
+  "speech_encoder_ds_rate": 5,
+  "speech_encoder_hidden_size": 1280,
+  "speech_encoder_type": "whisper",
+  "speech_normalize": false,
+  "speech_projector_lr": null,
+  "speech_projector_type": "linear",
+  "tie_word_embeddings": false,
+  "tokenizer_model_max_length": 8192,
+  "tokenizer_padding_side": "right",
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.44.0",
+  "use_cache": true,
+  "use_mm_proj": true,
+  "use_pos_skipping": false,
+  "use_sliding_window": false,
+  "vision_tower_pretrained": null,
+  "vocab_size": 152064
+}

generation_config.json

@@ -0,0 +1,16 @@
+{
+  "_attn_implementation": "flash_attention_2",
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.05,
+  "rope_theta": 1000000000.0,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "4.44.0"
+}

model-00001-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e9f66f1222d2e888f00cc70adc8dfc4101ed8ad38359ebd6cd0f3fd4b8614d9
+size 4877660776

model-00002-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c26fc99dfe1f1e8e381d51eb593060357decfbd4022d2b46f802195ce6225015
+size 4932751008

model-00003-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39fb6b27321e2f5c1c63342ecc5119cf42f664fd590e9837f5ef662420128141
+size 4998766592

model-00004-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d51535dfbcd23e1d19266528947966e01ae29060ba1ebec85d8aaeb49440d8c2
+size 2377117192

special_tokens_map.json

@@ -0,0 +1,20 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer_config.json

@@ -0,0 +1,44 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "model_max_length": 8192,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7474cc273e428e9606038894b8761d41ebce90cc8cf029160773d30686a31d2e
+size 7288