Upload folder using huggingface_hub (#1)

- a2841d09a9c43d55e6460f87c8ee75e22ca9ba4efd039d518306cd2b5b1e9971 (34fe5b9677d129a029cadebfa2225c6767454950)
- 7fac3e3d9eb926fc1106bb9830917b36085a6e9324e59dfe64219457ba86aa96 (b5aaa1e6a62a516dee037b6db22961314027c64c)

README.md

@@ -0,0 +1,85 @@
+---
+thumbnail: "https://assets-global.website-files.com/646b351987a8d8ce158d1940/64ec9e96b4334c0e1ac41504_Logo%20with%20white%20text.svg"
+base_model: tinyllava/TinyLLaVA-Phi-2-SigLIP-3.1B
+metrics:
+- memory_disk
+- memory_inference
+- inference_latency
+- inference_throughput
+- inference_CO2_emissions
+- inference_energy_consumption
+tags:
+- pruna-ai
+---
+<!-- header start -->
+<!-- 200823 -->
+<div style="width: auto; margin-left: auto; margin-right: auto">
+    <a href="https://www.pruna.ai/" target="_blank" rel="noopener noreferrer">
+        <img src="https://i.imgur.com/eDAlcgk.png" alt="PrunaAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
+    </a>
+</div>
+<!-- header end -->
+
+[![Twitter](https://img.shields.io/twitter/follow/PrunaAI?style=social)](https://twitter.com/PrunaAI)
+[![GitHub](https://img.shields.io/github/followers/PrunaAI?label=Follow%20%40PrunaAI&style=social)](https://github.com/PrunaAI)
+[![LinkedIn](https://img.shields.io/badge/LinkedIn-Connect-blue)](https://www.linkedin.com/company/93832878/admin/feed/posts/?feedType=following)
+[![Discord](https://img.shields.io/badge/Discord-Join%20Us-blue?style=social&logo=discord)](https://discord.gg/rskEr4BZJx)
+
+# Simply make AI models cheaper, smaller, faster, and greener!
+
+- Give a thumbs up if you like this model!
+- Contact us and tell us which model to compress next [here](https://www.pruna.ai/contact).
+- Request access to easily compress your *own* AI models [here](https://z0halsaff74.typeform.com/pruna-access?typeform-source=www.pruna.ai).
+- Read the documentations to know more [here](https://pruna-ai-pruna.readthedocs-hosted.com/en/latest/)
+- Join Pruna AI community on Discord [here](https://discord.gg/CP4VSgck) to share feedback/suggestions or get help.
+
+## Results
+
+![image info](./plots.png)
+
+**Frequently Asked Questions**
+- ***How does the compression work?*** The model is compressed with llm-int8.
+- ***How does the model quality change?*** The quality of the model output might vary compared to the base model.
+- ***How is the model efficiency evaluated?*** These results were obtained on HARDWARE_NAME with configuration described in `model/smash_config.json` and are obtained after a hardware warmup. The smashed model is directly compared to the original base model. Efficiency results may vary in other settings (e.g. other hardware, image size, batch size, ...). We recommend to directly run them in the use-case conditions to know if the smashed model can benefit you.
+- ***What is the model format?*** We use safetensors.
+- ***What calibration data has been used?*** If needed by the compression method, we used WikiText as the calibration data.
+- ***What is the naming convention for Pruna Huggingface models?*** We take the original model name and append "turbo", "tiny", or "green" if the smashed model has a measured inference speed, inference memory, or inference energy consumption which is less than 90% of the original base model.
+- ***How to compress my own models?*** You can request premium access to more compression methods and tech support for your specific use-cases [here](https://z0halsaff74.typeform.com/pruna-access?typeform-source=www.pruna.ai).
+- ***What are "first" metrics?*** Results mentioning "first" are obtained after the first run of the model. The first run might take more memory or be slower than the subsequent runs due cuda overheads.
+- ***What are "Sync" and "Async" metrics?*** "Sync" metrics are obtained by syncing all GPU processes and stop measurement when all of them are executed. "Async" metrics are obtained without syncing all GPU processes and stop when the model output can be used by the CPU. We provide both metrics since both could be relevant depending on the use-case. We recommend to test the efficiency gains directly in your use-cases.
+
+## Setup
+
+You can run the smashed model with these steps:
+
+0. Check requirements from the original repo tinyllava/TinyLLaVA-Phi-2-SigLIP-3.1B installed. In particular, check python, cuda, and transformers versions.
+1. Make sure that you have installed quantization related packages.
+    ```bash
+    pip install transformers accelerate bitsandbytes>0.37.0
+    ```
+2. Load & run the model.
+    ```python 
+   from transformers import AutoModelForCausalLM, AutoTokenizer
+   
+
+   model = AutoModelForCausalLM.from_pretrained("PrunaAI/tinyllava-TinyLLaVA-Phi-2-SigLIP-3.1B-bnb-4bit-smashed", trust_remote_code=True, device_map='auto')
+   tokenizer = AutoTokenizer.from_pretrained("tinyllava/TinyLLaVA-Phi-2-SigLIP-3.1B")
+    
+   input_ids = tokenizer("What is the color of prunes?,", return_tensors='pt').to(model.device)["input_ids"]
+    
+   outputs = model.generate(input_ids, max_new_tokens=216)
+   tokenizer.decode(outputs[0])
+    ```
+
+## Configurations
+
+The configuration info are in `smash_config.json`.
+
+## Credits & License
+
+The license of the smashed model follows the license of the original model. Please check the license of the original model tinyllava/TinyLLaVA-Phi-2-SigLIP-3.1B before using this model which provided the base model. The license  of the `pruna-engine` is [here](https://pypi.org/project/pruna-engine/) on Pypi.
+
+## Want to compress other models?
+
+- Contact us and tell us which model to compress next [here](https://www.pruna.ai/contact).
+- Request access to easily compress your own AI models [here](https://z0halsaff74.typeform.com/pruna-access?typeform-source=www.pruna.ai).

added_tokens.json

@@ -0,0 +1,40 @@
+{
+  "\t\t": 50294,
+  "\t\t\t": 50293,
+  "\t\t\t\t": 50292,
+  "\t\t\t\t\t": 50291,
+  "\t\t\t\t\t\t": 50290,
+  "\t\t\t\t\t\t\t": 50289,
+  "\t\t\t\t\t\t\t\t": 50288,
+  "\t\t\t\t\t\t\t\t\t": 50287,
+  "  ": 50286,
+  "   ": 50285,
+  "    ": 50284,
+  "     ": 50283,
+  "      ": 50282,
+  "       ": 50281,
+  "        ": 50280,
+  "         ": 50279,
+  "          ": 50278,
+  "           ": 50277,
+  "            ": 50276,
+  "             ": 50275,
+  "              ": 50274,
+  "               ": 50273,
+  "                ": 50272,
+  "                 ": 50271,
+  "                  ": 50270,
+  "                   ": 50269,
+  "                    ": 50268,
+  "                     ": 50267,
+  "                      ": 50266,
+  "                       ": 50265,
+  "                        ": 50264,
+  "                         ": 50263,
+  "                          ": 50262,
+  "                           ": 50261,
+  "                            ": 50260,
+  "                             ": 50259,
+  "                              ": 50258,
+  "                               ": 50257
+}

config.json

@@ -0,0 +1,93 @@
+{
+  "_name_or_path": "/ceph/hdd/staff/charpent/.cache/models2riyr4wgy2r4vryo",
+  "architectures": [
+    "TinyLlavaForConditionalGeneration"
+  ],
+  "auto_map": {
+    "AutoConfig": "configuration.TinyLlavaConfig",
+    "AutoModelForCausalLM": "modeling_tinyllava_phi.TinyLlavaForConditionalGeneration"
+  },
+  "cache_dir": null,
+  "connector_type": "mlp2x_gelu",
+  "hidden_size": 2560,
+  "ignore_index": -100,
+  "image_aspect_ratio": "square",
+  "image_token_index": -200,
+  "llm_model_name_or_path": "microsoft/phi-2",
+  "model_type": "tinyllava",
+  "num_queries": 128,
+  "num_resampler_layers": 3,
+  "pad_token": "<|endoftext|>",
+  "quantization_config": {
+    "_load_in_4bit": true,
+    "_load_in_8bit": false,
+    "bnb_4bit_compute_dtype": "bfloat16",
+    "bnb_4bit_quant_storage": "uint8",
+    "bnb_4bit_quant_type": "fp4",
+    "bnb_4bit_use_double_quant": false,
+    "llm_int8_enable_fp32_cpu_offload": false,
+    "llm_int8_has_fp16_weight": false,
+    "llm_int8_skip_modules": [
+      "lm_head"
+    ],
+    "llm_int8_threshold": 6.0,
+    "load_in_4bit": true,
+    "load_in_8bit": false,
+    "quant_method": "bitsandbytes"
+  },
+  "resampler_hidden_size": 768,
+  "text_config": {
+    "_name_or_path": "microsoft/phi-2",
+    "architectures": [
+      "PhiForCausalLM"
+    ],
+    "auto_map": {
+      "AutoConfig": "microsoft/phi-2--configuration_phi.PhiConfig",
+      "AutoModelForCausalLM": "microsoft/phi-2--modeling_phi.PhiForCausalLM"
+    },
+    "bos_token_id": 50256,
+    "embd_pdrop": 0.0,
+    "eos_token_id": 50256,
+    "hidden_act": "gelu_new",
+    "hidden_size": 2560,
+    "intermediate_size": 10240,
+    "layer_norm_eps": 1e-05,
+    "model_type": "phi",
+    "num_hidden_layers": 32,
+    "partial_rotary_factor": 0.4,
+    "qk_layernorm": false,
+    "resid_pdrop": 0.1,
+    "torch_dtype": "float16",
+    "vocab_size": 51200
+  },
+  "tokenizer_model_max_length": 3072,
+  "tokenizer_name_or_path": "microsoft/phi-2",
+  "tokenizer_padding_side": "right",
+  "tokenizer_use_fast": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.42.4",
+  "tune_type_connector": "full",
+  "tune_type_llm": "full",
+  "tune_type_vision_tower": "frozen",
+  "tune_vision_tower_from_layer": 0,
+  "use_cache": true,
+  "vision_config": {
+    "hidden_act": "gelu_pytorch_tanh",
+    "hidden_size": 1152,
+    "image_size": 384,
+    "intermediate_size": 4304,
+    "layer_norm_eps": 1e-06,
+    "model_name_or_path": "google/siglip-so400m-patch14-384",
+    "model_name_or_path2": "",
+    "model_type": "siglip_vision_model",
+    "num_attention_heads": 16,
+    "num_hidden_layers": 27,
+    "patch_size": 14
+  },
+  "vision_feature_layer": -2,
+  "vision_feature_select_strategy": "patch",
+  "vision_hidden_size": 1152,
+  "vision_model_name_or_path": "google/siglip-so400m-patch14-384",
+  "vision_model_name_or_path2": "",
+  "vocab_size": 51200
+}

configuration.py

@@ -0,0 +1,116 @@
+from transformers import PretrainedConfig
+from transformers import CONFIG_MAPPING
+from transformers import AutoConfig
+
+IGNORE_INDEX = -100
+IMAGE_TOKEN_INDEX = -200
+DEFAULT_IMAGE_TOKEN = "<image>"
+
+
+class TinyLlavaConfig(PretrainedConfig):
+
+    model_type = "tinyllava"
+    def __init__(
+        self,
+        llm_model_name_or_path = '',
+        tokenizer_name_or_path = None,
+        vision_model_name_or_path = '',
+        vision_model_name_or_path2 = '',
+        connector_type = None,
+        text_config=None,
+        hidden_size=2048,
+        vocab_size=32000,
+        ignore_index=-100,
+        image_token_index=32000,
+        pad_token = None,
+        pad_token_id = None,
+        tokenizer_padding_side = 'right',
+        tokenizer_model_max_length = 2048,
+        vision_config = None,
+        vision_hidden_size = None,
+        vision_feature_layer = -2,
+        vision_feature_select_strategy = 'patch',
+        image_aspect_ratio = 'square',
+        resampler_hidden_size = None,
+        num_queries = None,
+        num_resampler_layers = None,
+        use_cache = False,
+        cache_dir = None,
+        tokenizer_use_fast = False,
+        tune_type_llm = 'frozen',
+        tune_type_connector = 'frozen',
+        tune_type_vision_tower = 'frozen',
+        tune_vision_tower_from_layer = -1,
+        
+        **kwargs
+
+    ):
+        self.llm_model_name_or_path = llm_model_name_or_path
+        self.tokenizer_name_or_path = tokenizer_name_or_path or self.llm_model_name_or_path
+        self.vision_model_name_or_path = vision_model_name_or_path
+        self.vision_model_name_or_path2 = vision_model_name_or_path2
+        self.connector_type = connector_type
+        self.tune_type_llm = tune_type_llm
+        self.tune_type_connector = tune_type_connector
+        self.tune_type_vision_tower = tune_type_vision_tower
+        self.tune_vision_tower_from_layer = tune_vision_tower_from_layer
+        
+        self.ignore_index = IGNORE_INDEX
+        self.image_token_index = IMAGE_TOKEN_INDEX
+        self.pad_token = pad_token
+        self.pad_token_id = pad_token_id
+        self.tokenizer_padding_side = tokenizer_padding_side
+        self.tokenizer_model_max_length = tokenizer_model_max_length
+        self.vision_feature_layer = vision_feature_layer
+        self.vision_feature_select_strategy = vision_feature_select_strategy
+        self.image_aspect_ratio = image_aspect_ratio
+        self.resampler_hidden_size = resampler_hidden_size
+        self.num_queries = num_queries
+        self.num_resampler_layers = num_resampler_layers
+        self.use_cache = use_cache
+        self.cache_dir = cache_dir
+        self.tokenizer_use_fast = tokenizer_use_fast
+        self._load_text_config(text_config)
+        self._load_vision_config(vision_config)
+            
+        super().__init__(**kwargs)
+      
+    
+    def _load_text_config(self, text_config=None):
+        if self.llm_model_name_or_path is None or self.llm_model_name_or_path == '':
+            self.text_config = CONFIG_MAPPING['llama']()
+           
+        else:
+            self.text_config = AutoConfig.from_pretrained(self.llm_model_name_or_path, trust_remote_code=True)
+            if text_config is not None:
+                self.text_config = self.text_config.from_dict(text_config)
+                
+        self.hidden_size = getattr(self.text_config, 'hidden_size',  getattr(self.text_config, 'model_dim', None))
+        self.vocab_size = getattr(self.text_config, 'vocab_size',  None)
+    
+    
+    
+    def _load_vision_config(self, vision_config=None):
+        if self.vision_model_name_or_path is None or self.vision_model_name_or_path == '':
+            self.vision_config = CONFIG_MAPPING['clip_vision_model'](
+                intermediate_size=4096,
+                hidden_size=1024,
+                patch_size=14,
+                image_size=336,
+                num_hidden_layers=24,
+                num_attention_heads=16,
+                vocab_size=32000,
+                projection_dim=768,
+            )
+            
+        else:
+            self.vision_config = AutoConfig.from_pretrained(self.vision_model_name_or_path.split(':')[-1])
+            self.vision_config = getattr(self.vision_config, 'vision_config', self.vision_config)
+            if vision_config is not None:
+                self.vision_config = self.vision_config.from_dict(vision_config)
+                
+        self.vision_config.model_name_or_path = self.vision_model_name_or_path.split(':')[-1]
+        self.vision_config.model_name_or_path2 = self.vision_model_name_or_path2.split(':')[-1]
+        self.vision_hidden_size = getattr(self.vision_config, 'hidden_size',  None)  
+        
+

generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 50256,
+  "eos_token_id": 50256,
+  "transformers_version": "4.42.4",
+  "use_cache": false
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41a8f215af9be6e29660e87383bce1d2163b762b226bb5c75e943964ea10dc6d
+size 2196845887

modeling_tinyllava_phi.py

@@ -0,0 +1,624 @@
+# For licensing see accompanying LICENSE file.
+# Copyright (C) 2024 TinyLLaVA. All Rights Reserved.
+import time
+
+import dataclasses
+from enum import auto, Enum
+from typing import List, Tuple, Optional, Union
+import requests
+from PIL import Image
+from io import BytesIO
+import base64
+import re
+
+import torch
+import torch.utils.checkpoint
+from torch import nn
+from torch.nn import functional as F
+
+from transformers.utils import logging
+from transformers import PreTrainedModel
+from transformers.modeling_outputs import CausalLMOutputWithPast
+from transformers.generation.utils import GenerateOutput
+from transformers import CLIPVisionModel, CLIPImageProcessor, SiglipVisionModel, SiglipImageProcessor
+
+from .configuration import TinyLlavaConfig, IGNORE_INDEX, IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN
+
+from transformers import AutoConfig, AutoModelForCausalLM, PhiForCausalLM
+
+
+
+logger = logging.get_logger(__name__)
+
+# Model Constants
+IGNORE_INDEX = -100
+IMAGE_TOKEN_INDEX = -200
+DEFAULT_IMAGE_TOKEN = "<image>"
+DEFAULT_IMAGE_PATCH_TOKEN = "<im_patch>"
+DEFAULT_IM_START_TOKEN = "<im_start>"
+DEFAULT_IM_END_TOKEN = "<im_end>"
+IMAGE_PLACEHOLDER = "<image-placeholder>"
+
+CONTROLLER_HEART_BEAT_EXPIRATION = 30
+WORKER_HEART_BEAT_INTERVAL = 15
+LOGDIR = "."
+
+
+class SeparatorStyle(Enum):
+    """Different separator style."""
+    SINGLE = auto()
+    TWO = auto()
+    MPT = auto()
+    PLAIN = auto()
+    LLAMA_2 = auto()
+    TINY_LLAMA = auto()
+    QWEN_2 = auto()
+
+
+@dataclasses.dataclass
+class Conversation:
+    """A class that keeps all conversation history."""
+    system: str
+    roles: List[str]
+    messages: List[List[str]]
+    offset: int
+    sep_style: SeparatorStyle = SeparatorStyle.SINGLE
+    sep: str = "###"
+    sep2: str = None
+    version: str = "Unknown"
+
+    skip_next: bool = False
+
+    def get_prompt(self):
+        messages = self.messages
+        if len(messages) > 0 and type(messages[0][1]) is tuple:
+            messages = self.messages.copy()
+            init_role, init_msg = messages[0].copy()
+            init_msg = init_msg[0].replace("<image>", "").strip()
+            if 'mmtag' in self.version:
+                messages[0] = (init_role, init_msg)
+                messages.insert(0, (self.roles[0], "<Image><image></Image>"))
+                messages.insert(1, (self.roles[1], "Received."))
+            else:
+                messages[0] = (init_role, "<image>\n" + init_msg)
+
+        if self.sep_style == SeparatorStyle.TWO:
+            seps = [self.sep, self.sep2]
+            ret = self.system + seps[0]
+            for i, (role, message) in enumerate(messages):
+                if message:
+                    if type(message) is tuple:
+                        message, _, _ = message
+                    ret += role + ": " + message + seps[i % 2]
+                else:
+                    ret += role + ":"
+        else:
+            raise ValueError(f"Invalid style: {self.sep_style}")
+
+        return ret
+
+    def append_message(self, role, message):
+        self.messages.append([role, message])
+
+    def copy(self):
+        return Conversation(
+            system=self.system,
+            roles=self.roles,
+            messages=[[x, y] for x, y in self.messages],
+            offset=self.offset,
+            sep_style=self.sep_style,
+            sep=self.sep,
+            sep2=self.sep2,
+            version=self.version)
+
+
+
+
+conv_phi_v0 = Conversation(
+    system="A chat between a curious user and an artificial intelligence assistant. "
+           "The assistant gives helpful, detailed, and polite answers to the user's questions.",
+    roles=("USER", "ASSISTANT"),
+    version="phi",
+    messages=(),
+    offset=0,
+    sep_style=SeparatorStyle.TWO,
+    sep=" ",
+    sep2="<|endoftext|>",
+)
+
+
+def load_image_from_base64(image):
+    return Image.open(BytesIO(base64.b64decode(image)))
+
+
+def expand2square(pil_img, background_color):
+    width, height = pil_img.size
+    if width == height:
+        return pil_img
+    elif width > height:
+        result = Image.new(pil_img.mode, (width, width), background_color)
+        result.paste(pil_img, (0, (width - height) // 2))
+        return result
+    else:
+        result = Image.new(pil_img.mode, (height, height), background_color)
+        result.paste(pil_img, ((height - width) // 2, 0))
+        return result
+
+
+def process_images(images, image_processor, model_cfg):
+    image_aspect_ratio = getattr(model_cfg, "image_aspect_ratio", None)
+    new_images = []
+    if image_aspect_ratio == 'pad':
+        for image in images:
+            image = expand2square(image, tuple(int(x*255) for x in image_processor.image_mean))
+            image = image_processor.preprocess(image, return_tensors='pt')['pixel_values'][0]
+            new_images.append(image)
+    else:
+        return image_processor(images, return_tensors='pt')['pixel_values']
+    if all(x.shape == new_images[0].shape for x in new_images):
+        new_images = torch.stack(new_images, dim=0)
+    return new_images
+
+
+def tokenizer_image_token(prompt, tokenizer, image_token_index=IMAGE_TOKEN_INDEX, return_tensors=None):
+    prompt_chunks = [tokenizer(chunk).input_ids for chunk in prompt.split('<image>')]
+
+    def insert_separator(X, sep):
+        return [ele for sublist in zip(X, [sep]*len(X)) for ele in sublist][:-1]
+
+    input_ids = []
+    offset = 0
+    if len(prompt_chunks) > 0 and len(prompt_chunks[0]) > 0 and prompt_chunks[0][0] == tokenizer.bos_token_id:
+        offset = 1
+        input_ids.append(prompt_chunks[0][0])
+
+    for x in insert_separator(prompt_chunks, [image_token_index] * (offset + 1)):
+        input_ids.extend(x[offset:])
+
+    if return_tensors is not None:
+        if return_tensors == 'pt':
+            return torch.tensor(input_ids, dtype=torch.long)
+        raise ValueError(f'Unsupported tensor type: {return_tensors}')
+    return input_ids
+
+def load_image(image_file):
+    if image_file.startswith("http") or image_file.startswith("https"):
+        response = requests.get(image_file)
+        image = Image.open(BytesIO(response.content)).convert("RGB")
+    else:
+        image = Image.open(image_file).convert("RGB")
+    return image
+
+ACT_TYPE = {
+    'relu': nn.ReLU,
+    'gelu': nn.GELU
+}
+
+class Connector(nn.Module):
+    def __init__(self, config=None):
+        super().__init__()
+        mlp_gelu_match = re.match(r'^mlp(\d+)x_gelu$', config.connector_type)
+        act_type = config.connector_type.split('_')[-1]
+        mlp_depth = int(mlp_gelu_match.group(1))
+        modules = [nn.Linear(config.vision_hidden_size, config.hidden_size)]
+        for _ in range(1, mlp_depth):
+            modules.append(ACT_TYPE[act_type]())
+            modules.append(nn.Linear(config.hidden_size, config.hidden_size))
+            
+        self._connector = nn.Sequential(*modules)
+    
+    def forward(self, x):
+        return self._connector(x)
+
+class VisionTower(nn.Module):
+    def __init__(self, cfg, model_name_or_path = 'clip'):
+        super().__init__()
+        if 'clip' in model_name_or_path:
+            self._vision_tower = CLIPVisionModel(cfg)
+            self._image_processor = CLIPImageProcessor.from_pretrained(cfg.model_name_or_path)
+        else:
+            self._vision_tower = SiglipVisionModel(cfg)
+            self._image_processor = SiglipImageProcessor.from_pretrained(cfg.model_name_or_path)
+            
+        self.config = cfg
+        
+    def forward(self, x, **kwargs):
+        image_features = self._vision_tower(x, output_hidden_states=True)
+        image_features = image_features.hidden_states[kwargs.get('vision_feature_layer', -2)]
+
+        if kwargs.get('vision_feature_select_strategy', 'patch') == 'patch':
+            image_features = image_features[:, 1:]
+        elif kwargs.get('vision_feature_select_strategy', 'patch') == 'cls_patch':
+            image_features = image_features
+        else:
+            raise ValueError(f"Unexpected select feature: {kwargs.get('vision_feature_select_strategy')}")
+
+        return image_features
+        
+    @property
+    def vision_tower(self):
+        return self._vision_tower
+        
+    @vision_tower.setter
+    def vision_tower(self, vision_tower):
+        self._vision_tower = vision_tower
+
+def get_value_from_kwargs(kwargs, name):
+    if name in kwargs:
+        return kwargs.pop(name)
+    else:
+        return None
+    
+
+class TinyLlavaPreTrainedModel(PreTrainedModel):
+    config_class = TinyLlavaConfig
+    base_model_prefix = "model"
+    supports_gradient_checkpointing = True
+    _no_split_modules = ["LlavaVisionAttention"]
+    _skip_keys_device_placement = "past_key_values"
+    _supports_flash_attn_2 = True
+
+    def _init_weights(self, module):
+        std = (
+            self.config.initializer_range
+            if hasattr(self.config, "initializer_range")
+            else self.config.text_config.initializer_range
+        )
+
+        if hasattr(module, "class_embedding"):
+            module.class_embedding.data.normal_(mean=0.0, std=std)
+
+        if isinstance(module, (nn.Linear, nn.Conv2d)):
+            module.weight.data.normal_(mean=0.0, std=std)
+            if module.bias is not None:
+                module.bias.data.zero_()
+        elif isinstance(module, nn.Embedding):
+            module.weight.data.normal_(mean=0.0, std=std)
+            if module.padding_idx is not None:
+                module.weight.data[module.padding_idx].zero_()
+
+    @property
+    def _supports_sdpa(self):
+        return self.language_model._supports_sdpa
+
+
+class TinyLlavaForConditionalGeneration(TinyLlavaPreTrainedModel):
+    def __init__(self, config: TinyLlavaConfig):
+        
+        super().__init__(config)
+
+        self.language_model = PhiForCausalLM(config.text_config)
+        self.vision_tower = VisionTower(config.vision_config, config.vision_model_name_or_path)
+        self.connector = Connector(config)
+        self.post_init()
+
+    
+    def get_input_embeddings(self):
+        return self.language_model.get_input_embeddings()
+
+    def set_input_embeddings(self, value):
+        self.language_model.set_input_embeddings(value)
+
+    def get_output_embeddings(self):
+        return self.language_model.get_output_embeddings()
+
+    def set_output_embeddings(self, new_embeddings):
+        self.language_model.set_output_embeddings(new_embeddings)
+
+    def set_decoder(self, decoder):
+        self.language_model.set_decoder(decoder)
+
+    def get_decoder(self):
+        return self.language_model.get_decoder()
+
+    def tie_weights(self):
+        return self.language_model.tie_weights()
+
+    def resize_token_embeddings(self, new_num_tokens: Optional[int] = None, pad_to_multiple_of=None) -> nn.Embedding:
+        model_embeds = self.language_model.resize_token_embeddings(new_num_tokens, pad_to_multiple_of)
+        # update vocab size
+        self.config.text_config.vocab_size = model_embeds.num_embeddings
+        self.config.vocab_size = model_embeds.num_embeddings
+        self.vocab_size = model_embeds.num_embeddings
+        return model_embeds
+
+    
+    def forward(
+        self,
+        input_ids: torch.LongTensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[List[torch.FloatTensor]] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        labels: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        images: Optional[torch.FloatTensor] = None,
+        image_sizes: Optional[List[List[int]]] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, CausalLMOutputWithPast]:
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+        if inputs_embeds is None:
+            (
+                input_ids,
+                position_ids,
+                attention_mask,
+                past_key_values,
+                inputs_embeds,
+                labels
+            ) = self.prepare_inputs_labels_for_multimodal(
+                input_ids,
+                position_ids,
+                attention_mask,
+                past_key_values,
+                labels,
+                images,
+                image_sizes
+            )
+        return self.language_model.forward(
+            input_ids=input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            labels=labels,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict
+        )
+    
+    @torch.no_grad()
+    def generate(
+        self,
+        inputs: Optional[torch.Tensor] = None,
+        images: Optional[torch.Tensor] = None,
+        image_sizes: Optional[torch.Tensor] = None,
+        **kwargs,
+    ) -> Union[GenerateOutput, torch.LongTensor]:
+        position_ids = kwargs.pop("position_ids", None)
+        attention_mask = kwargs.pop("attention_mask", None)
+        if "inputs_embeds" in kwargs:
+            raise NotImplementedError("`inputs_embeds` is not supported")
+
+        if images is not None:
+            (
+                inputs,
+                position_ids,
+                attention_mask,
+                _,
+                inputs_embeds,
+                _
+            ) = self.prepare_inputs_labels_for_multimodal(
+                inputs,
+                position_ids,
+                attention_mask,
+                None,
+                None,
+                images,
+                image_sizes=image_sizes
+            )
+        else:
+            inputs_embeds = self.language_model.get_input_embeddings()(inputs)
+
+        return self.language_model.generate(
+            position_ids=position_ids,
+            attention_mask=attention_mask,
+            inputs_embeds=inputs_embeds,
+            **kwargs
+        )
+        
+    def encode_images(self, images):
+        kwargs = {}
+        kwargs['vision_feature_layer'] = self.config.vision_feature_layer
+        kwargs['vision_feature_select_strategy'] = self.config.vision_feature_select_strategy
+        images = images.to(device=self.device, dtype=self.dtype)
+        image_features = self.vision_tower(images, **kwargs)
+        image_features = self.connector(image_features)
+        return image_features
+    
+    
+    
+    def prepare_inputs_for_generation(self, input_ids, past_key_values=None,
+                                      inputs_embeds=None, **kwargs):
+        images = kwargs.pop("images", None)
+        image_sizes = kwargs.pop("image_sizes", None)
+        inputs = self.language_model.prepare_inputs_for_generation(
+            input_ids, past_key_values=past_key_values, inputs_embeds=inputs_embeds, **kwargs
+        )
+        if images is not None:
+            inputs['images'] = images
+        if image_sizes is not None:
+            inputs['image_sizes'] = image_sizes
+        return inputs
+        
+    def prepare_inputs_labels_for_multimodal(
+        self, input_ids, position_ids, attention_mask, past_key_values, labels,
+        images, image_sizes=None
+    ):
+        vision_tower = self.vision_tower
+        if vision_tower is None or images is None or input_ids.shape[1] == 1:
+            return input_ids, position_ids, attention_mask, past_key_values, None, labels
+
+        
+        image_features = self.encode_images(images)
+
+        # TODO: image start / end is not implemented here to support pretraining.
+        if getattr(self.config, 'tune_mm_mlp_adapter', False):
+            raise NotImplementedError
+
+        # Let's just add dummy tensors if they do not exist,
+        # it is a headache to deal with None all the time.
+        # But it is not ideal, and if you have a better idea,
+        # please open an issue / submit a PR, thanks.
+        _labels = labels
+        _position_ids = position_ids
+        _attention_mask = attention_mask
+        if attention_mask is None:
+            attention_mask = torch.ones_like(input_ids, dtype=torch.bool)
+        else:
+            attention_mask = attention_mask.bool()
+        if position_ids is None:
+            position_ids = torch.arange(0, input_ids.shape[1], dtype=torch.long, device=input_ids.device)
+        if labels is None:
+            labels = torch.full_like(input_ids, IGNORE_INDEX)
+
+        # remove the padding using attention_mask -- FIXME
+        _input_ids = input_ids
+        input_ids = [cur_input_ids[cur_attention_mask] for cur_input_ids, cur_attention_mask in zip(input_ids, attention_mask)]
+        labels = [cur_labels[cur_attention_mask] for cur_labels, cur_attention_mask in zip(labels, attention_mask)]
+
+        new_input_embeds = []
+        new_labels = []
+        cur_image_idx = 0
+        for batch_idx, cur_input_ids in enumerate(input_ids):
+            num_images = (cur_input_ids == IMAGE_TOKEN_INDEX).sum()
+            if num_images == 0:
+                cur_image_features = image_features[cur_image_idx]
+                cur_input_embeds_1 = self.language_model.get_input_embeddings()(cur_input_ids)
+                cur_input_embeds = torch.cat([cur_input_embeds_1, cur_image_features[0:0]], dim=0)
+                new_input_embeds.append(cur_input_embeds)
+                new_labels.append(labels[batch_idx])
+                cur_image_idx += 1
+                continue
+
+            image_token_indices = [-1] + torch.where(cur_input_ids == IMAGE_TOKEN_INDEX)[0].tolist() + [cur_input_ids.shape[0]]
+            cur_input_ids_noim = []
+            cur_labels = labels[batch_idx]
+            cur_labels_noim = []
+            for i in range(len(image_token_indices) - 1):
+                cur_input_ids_noim.append(cur_input_ids[image_token_indices[i]+1:image_token_indices[i+1]])
+                cur_labels_noim.append(cur_labels[image_token_indices[i]+1:image_token_indices[i+1]])
+            split_sizes = [x.shape[0] for x in cur_labels_noim]
+            cur_input_embeds = self.language_model.get_input_embeddings()(torch.cat(cur_input_ids_noim))
+            cur_input_embeds_no_im = torch.split(cur_input_embeds, split_sizes, dim=0)
+            cur_new_input_embeds = []
+            cur_new_labels = []
+
+            for i in range(num_images + 1):
+                cur_new_input_embeds.append(cur_input_embeds_no_im[i])
+                cur_new_labels.append(cur_labels_noim[i])
+                if i < num_images:
+                    cur_image_features = image_features[cur_image_idx]
+                    cur_image_idx += 1
+                    cur_new_input_embeds.append(cur_image_features)
+                    cur_new_labels.append(torch.full((cur_image_features.shape[0],), IGNORE_INDEX, device=cur_labels.device, dtype=cur_labels.dtype))
+
+            cur_new_input_embeds = [x.to(self.device) for x in cur_new_input_embeds]
+
+            cur_new_input_embeds = torch.cat(cur_new_input_embeds)
+            cur_new_labels = torch.cat(cur_new_labels)
+
+            new_input_embeds.append(cur_new_input_embeds)
+            new_labels.append(cur_new_labels)
+
+        # Truncate sequences to max length as image embeddings can make the sequence longer
+        tokenizer_model_max_length = getattr(self.config, 'tokenizer_model_max_length', None)
+        if tokenizer_model_max_length is not None:
+            new_input_embeds = [x[:tokenizer_model_max_length] for x in new_input_embeds]
+            new_labels = [x[:tokenizer_model_max_length] for x in new_labels]
+
+        # Combine them
+        max_len = max(x.shape[0] for x in new_input_embeds)
+        batch_size = len(new_input_embeds)
+
+        new_input_embeds_padded = []
+        new_labels_padded = torch.full((batch_size, max_len), IGNORE_INDEX, dtype=new_labels[0].dtype, device=new_labels[0].device)
+        attention_mask = torch.zeros((batch_size, max_len), dtype=attention_mask.dtype, device=attention_mask.device)
+        position_ids = torch.zeros((batch_size, max_len), dtype=position_ids.dtype, device=position_ids.device)
+
+        for i, (cur_new_embed, cur_new_labels) in enumerate(zip(new_input_embeds, new_labels)):
+            cur_len = cur_new_embed.shape[0]
+            if getattr(self.config, 'tokenizer_padding_side', 'right') == "left":
+                new_input_embeds_padded.append(torch.cat((
+                    torch.zeros((max_len - cur_len, cur_new_embed.shape[1]), dtype=cur_new_embed.dtype, device=cur_new_embed.device),
+                    cur_new_embed
+                ), dim=0))
+                if cur_len > 0:
+                    new_labels_padded[i, -cur_len:] = cur_new_labels
+                    attention_mask[i, -cur_len:] = True
+                    position_ids[i, -cur_len:] = torch.arange(0, cur_len, dtype=position_ids.dtype, device=position_ids.device)
+            else:
+                new_input_embeds_padded.append(torch.cat((
+                    cur_new_embed,
+                    torch.zeros((max_len - cur_len, cur_new_embed.shape[1]), dtype=cur_new_embed.dtype, device=cur_new_embed.device)
+                ), dim=0))
+                if cur_len > 0:
+                    new_labels_padded[i, :cur_len] = cur_new_labels
+                    attention_mask[i, :cur_len] = True
+                    position_ids[i, :cur_len] = torch.arange(0, cur_len, dtype=position_ids.dtype, device=position_ids.device)
+
+        new_input_embeds = torch.stack(new_input_embeds_padded, dim=0)
+
+        if _labels is None:
+            new_labels = None
+        else:
+            new_labels = new_labels_padded
+
+        if _attention_mask is None:
+            attention_mask = None
+        else:
+            attention_mask = attention_mask.to(dtype=_attention_mask.dtype)
+
+        if _position_ids is None:
+            position_ids = None
+
+        return None, position_ids, attention_mask, past_key_values, new_input_embeds, new_labels
+    
+    def chat(
+        self,
+        prompt: str,
+        tokenizer = None,
+        image: str = None,
+        max_new_tokens: int = 512,
+        num_beams = 1,
+        top_p=None,
+        temperature=0
+    ):
+        image_processor = self.vision_tower._image_processor
+
+        if image is not None:
+            prompt = DEFAULT_IMAGE_TOKEN + '\n' + prompt 
+        conv = conv_phi_v0.copy()
+        conv.append_message(conv.roles[0], prompt)
+        conv.append_message(conv.roles[1], None)
+        prompt = conv.get_prompt()
+        if image is not None:
+            image = load_image(image)
+            image_tensor = process_images(image, image_processor, self.config).to(self.device)
+
+        input_ids = (
+            tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors="pt")
+            .unsqueeze(0).to(self.device)
+        )
+        # Generate
+        stime = time.time()
+
+        with torch.inference_mode():
+            output_ids = self.generate(
+                input_ids,
+                images=image_tensor,
+                do_sample=True if temperature > 0 else False,
+                temperature=temperature,
+                top_p=top_p,
+                num_beams=num_beams,
+                pad_token_id=tokenizer.pad_token_id,
+                max_new_tokens=max_new_tokens,
+                use_cache=True,
+                # stopping_criteria=[stopping_criteria],
+            )
+
+        # print('inference over')
+        generation_time = time.time() - stime
+        outputs = tokenizer.batch_decode(
+            output_ids, skip_special_tokens=True
+        )[0]
+
+        outputs = outputs.strip()
+
+        return outputs, generation_time
+    
+
+AutoConfig.register("tinyllava", TinyLlavaConfig)        
+AutoModelForCausalLM.register(TinyLlavaConfig, TinyLlavaForConditionalGeneration)

smash_config.json

@@ -0,0 +1,31 @@
+{
+    "api_key": null,
+    "verify_url": "http://johnrachwan.pythonanywhere.com",
+    "smash_config": {
+        "pruners": "None",
+        "pruning_ratio": 0.0,
+        "factorizers": "None",
+        "quantizers": "['llm-int8']",
+        "weight_quantization_bits": 4,
+        "output_deviation": 0.005,
+        "compilers": "None",
+        "static_batch": true,
+        "static_shape": true,
+        "controlnet": "None",
+        "unet_dim": 4,
+        "device": "cuda",
+        "cache_dir": "/ceph/hdd/staff/charpent/.cache/models2riyr4wg",
+        "batch_size": 1,
+        "model_name": "tinyllava/TinyLLaVA-Phi-2-SigLIP-3.1B",
+        "task": "text_text_generation",
+        "max_batch_size": 1,
+        "qtype_weight": "torch.qint8",
+        "qtype_activation": "torch.quint8",
+        "qobserver": "<class 'torch.ao.quantization.observer.MinMaxObserver'>",
+        "qscheme": "torch.per_tensor_symmetric",
+        "qconfig": "x86",
+        "group_size": 128,
+        "damp_percent": 0.1,
+        "save_load_fn": "bitsandbytes"
+    }
+}

special_tokens_map.json

@@ -0,0 +1,30 @@
+{
+  "bos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer_config.json

@@ -0,0 +1,329 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "50256": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50257": {
+      "content": "                               ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50258": {
+      "content": "                              ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50259": {
+      "content": "                             ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50260": {
+      "content": "                            ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50261": {
+      "content": "                           ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50262": {
+      "content": "                          ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50263": {
+      "content": "                         ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50264": {
+      "content": "                        ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50265": {
+      "content": "                       ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50266": {
+      "content": "                      ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50267": {
+      "content": "                     ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50268": {
+      "content": "                    ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50269": {
+      "content": "                   ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50270": {
+      "content": "                  ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50271": {
+      "content": "                 ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50272": {
+      "content": "                ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50273": {
+      "content": "               ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50274": {
+      "content": "              ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50275": {
+      "content": "             ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50276": {
+      "content": "            ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50277": {
+      "content": "           ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50278": {
+      "content": "          ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50279": {
+      "content": "         ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50280": {
+      "content": "        ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50281": {
+      "content": "       ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50282": {
+      "content": "      ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50283": {
+      "content": "     ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50284": {
+      "content": "    ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50285": {
+      "content": "   ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50286": {
+      "content": "  ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50287": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50288": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50289": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50290": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50291": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50292": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50293": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50294": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "bos_token": "<|endoftext|>",
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "legacy": false,
+  "model_max_length": 3072,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "return_token_type_ids": false,
+  "tokenizer_class": "CodeGenTokenizer",
+  "unk_token": "<|endoftext|>"
+}