backup

modeling_cocom.py

@@ -1,34 +1,32 @@
 from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig, PreTrainedModel, PretrainedConfig, AutoModel
-from linformer.attention import LinformerSelfAttention
 import torch
 import math 
 from peft import get_peft_model, LoraConfig, TaskType
 import os
 
-
-# Freeze model function (unchanged)
 def freeze_model(model):
     for param in model.parameters():
         param.requires_grad = False
 
 
-# BERT_Compressor remains the same as you are not modifying it for Linformer
 class BERT_Compressor(torch.nn.Module):
     def __init__(self, compr_model_name, compr_rate, compr_linear_type, decoder_hidden_size):
         super().__init__()
-        self.model_name = compr_model_name
+        # init model
+        self.model_name = compr_model_name # base model name of BERT; example: bert-base-ucased
         self.model = AutoModel.from_pretrained(compr_model_name, torch_dtype=torch.float16)
         self.tokenizer = AutoTokenizer.from_pretrained(compr_model_name, use_fast=True) 
-        self.compr_rate = compr_rate
-        self.compressing_mode = compr_linear_type
+        self.compr_rate = compr_rate # compression rate
+        self.compressing_mode = compr_linear_type # linear layer type, could be either concat or mean.
 
-        if self.compressing_mode == 'concat':
+        if self.compressing_mode == 'concat': # default setting in paper
             self.linear = torch.nn.Linear(self.model.config.hidden_size*self.compr_rate, decoder_hidden_size) 
         elif self.compressing_mode == 'mean':
             self.linear = torch.nn.Linear(self.model.config.hidden_size, decoder_hidden_size)
         self.linear = self.linear.float16()
 
     def forward(self, input_ids, attention_mask):
+        # compressing context using BERT
         segment_compress_outputs = self.model(input_ids=input_ids, attention_mask=attention_mask, output_hidden_states=True) 
         num_embs = math.ceil(input_ids.size(1) / self.compr_rate)
         all_hidden_states_emb = list()
@@ -44,18 +42,23 @@ class BERT_Compressor(torch.nn.Module):
                 start_idx = segment_idx * self.compr_rate
                 end_idx = (segment_idx + 1) * self.compr_rate
                 hidden_state = segment_compress_outputs.hidden_states[-1][:, start_idx:end_idx, :]
+                # Apply mean pooling to get the final embedding for the segment
                 all_hidden_states_emb.append(hidden_state)
+        else: 
+            raise NotImplementedError()
+        
         all_hidden_states_emb_cat = torch.stack(all_hidden_states_emb, dim=1)
         transformed_embeds = self.linear(all_hidden_states_emb_cat)
         
+
         if self.compressing_mode == "mean":
             transformed_embeds = torch.mean(transformed_embeds, dim=2)
 
+        # dimention of transformed_embeds: (batch_size*generation_top_k, num_embs, decoder_hidden_size)
         return  transformed_embeds
 
-
-# Modify COCOMConfig to support Linformer
 class COCOMConfig(PretrainedConfig):
+
     model_type = "COCOM"
     def __init__(self,
                 decoder_model_name="meta-llama/Llama-2-7b-chat-hf",
@@ -68,78 +71,189 @@ class COCOMConfig(PretrainedConfig):
                 lora = False,
                 training_form="both",
                 lora_r=16,
-                attn_implementation="linformer",  # Change default to Linformer
+                attn_implementation="eager",
                 device_map = "cuda",
                  **kwargs):
         super().__init__(**kwargs)
-        self.decoder_model_name = decoder_model_name 
-        self.quantization = quantization 
-        self.generation_top_k = generation_top_k 
-        self.sep = sep 
-        self.compr_model_name = compr_model_name 
-        self.compr_rate = compr_rate 
-        self.compr_linear_type = compr_linear_type 
-        self.lora = lora 
-        self.training_form = training_form 
-        self.lora_r = lora_r 
+
+        self.decoder_model_name = decoder_model_name # model name of decoder
+        self.quantization = quantization # quantization, could be no, int4, int8
+        self.generation_top_k = generation_top_k # top k for each query, for pretraining, set to 1
+        self.sep = sep # boolean type, whether to use sep token
+        self.compr_model_name = compr_model_name # model name of compressor
+        self.compr_rate = compr_rate # compression rate
+        self.compr_linear_type = compr_linear_type # linear layer type, could be either concat or mean
+        self.lora = lora # boolean type, whether to use lora trsining
+        self.training_form = training_form # training form, could be compressor: training only comprssor; both: 
+        self.lora_r = lora_r # lora_r for lora training, we use 16 throughout the experiment.
         self.attn_implementation = attn_implementation
         self.device_map = device_map
 
-
-# Modify COCOM model to use Linformer in the attention layer
 class COCOM(PreTrainedModel):
     config_class = COCOMConfig
     def __init__(self, cfg):
         super().__init__(cfg)
+        # define models
         attn_impl = cfg.attn_implementation
-
-        # Load the model (decoder) in standard quantization or Linformer
-        self.decoder = AutoModelForCausalLM.from_pretrained(
-            cfg.decoder_model_name, 
-            torch_dtype=torch.float16,
-            low_cpu_mem_usage=True,
-            device_map=cfg.device_map
-        )
+        # model could be loaded in three quantization modes: no, int4, int8
+        if cfg.quantization == "no":
+            self.decoder = AutoModelForCausalLM.from_pretrained(
+                cfg.decoder_model_name, 
+                torch_dtype=torch.float16,
+                attn_implementation=attn_impl, 
+                low_cpu_mem_usage = True,
+                device_map =cfg.device_map
+                )
+        elif cfg.quantization == "int4":
+            quant_config = BitsAndBytesConfig(
+                load_in_4bit=True,
+                bnb_4bit_quant_type='nf4',
+                bnb_4bit_compute_dtype='float16',
+                low_cpu_mem_usage = True,
+            )
+            self.decoder = AutoModelForCausalLM.from_pretrained(
+                cfg.decoder_model_name, 
+                quantization_config=quant_config,
+                attn_implementation=attn_impl, 
+                torch_dtype=torch.float16,
+                resume_download=True,
+                low_cpu_mem_usage = True,
+                trust_remote_code=True,
+                device_map =cfg.device_map
+            )
+        elif cfg.quantization == "int8":
+            quant_config = BitsAndBytesConfig(
+                load_in_8bit=True,
+                llm_int8_enable_fp32_cpu_offload=True,
+                bnb_4bit_compute_dtype='float16',
+                low_cpu_mem_usage = True,
+            )
+            self.decoder = AutoModelForCausalLM.from_pretrained(
+                cfg.decoder_model_name,
+                quantization_config=quant_config,
+                attn_implementation=attn_impl,
+                torch_dtype=torch.float16,
+                resume_download=True,
+                low_cpu_mem_usage = True,
+                trust_remote_code=True,
+                device_map =cfg.device_map
+            )
+        else:
+            raise NotImplementedError()
         
-        # Replace decoder's attention mechanism with LinformerSelfAttention if configured
-        if attn_impl == 'linformer':
-            self._replace_attention_with_linformer()
+        # when compr_model_name is not set, then means using a decoder-based compressor, otherwise a bert based compressor
+        if cfg.compr_model_name is not None:
+            # case bert based compressor
+            self.compr = BERT_Compressor(cfg.compr_model_name, cfg.compr_rate, cfg.compr_linear_type, self.decoder.config.hidden_size)
+        else:
+            # case decoder based compressor
+            self.compr = None
 
-        # Initialize other parts of the model (compression, LoRA, etc.)
-        self.compr = BERT_Compressor(cfg.compr_model_name, cfg.compr_rate, cfg.compr_linear_type, self.decoder.config.hidden_size)
+        # set lora adaptors
         if cfg.lora:
-            self._apply_lora(cfg.lora_r)
+            peft_config = LoraConfig(
+                        task_type="CAUSAL_LM",
+                        r=cfg.lora_r,
+                        lora_alpha=2* cfg.lora_r,
+                        target_modules='all-linear',
+                        lora_dropout=0.1,
+                    )
+            self.decoder = get_peft_model(self.decoder, peft_config)
+            self.decoder.print_trainable_parameters()  
+
+        # for training_form=compressor, then freeze the decoder for BERT-based
+        self.training_form = cfg.training_form
+        if self.training_form == "compressor" and self.compr is not None:
+            freeze_model(self.decoder)
 
         self.decoder_tokenizer = AutoTokenizer.from_pretrained(cfg.decoder_model_name, use_fast=True, padding_side='left')
+
+        # define special tokens
         self.decoder_tokenizer.add_special_tokens({'additional_special_tokens': ['<MEM>', '<AE>', '<ENC>', '<SEP>']})
+        self.decoder_tokenizer.mem_token = '<MEM>' # Memory token
+        self.decoder_tokenizer.ae_token = '<AE>' # token for autoencoding on decoder side
+        self.decoder_tokenizer.enc_token = '<ENC>' # token for autoencoding on compressor side
+        self.decoder_tokenizer.sep_token = '<SEP>' # sep token between document
+
+        self.decoder_tokenizer.mem_token_id = self.decoder_tokenizer.convert_tokens_to_ids('<MEM>')
+        self.decoder_tokenizer.ae_token_id = self.decoder_tokenizer.convert_tokens_to_ids('<AE>')
+        self.decoder_tokenizer.sep_token_id = self.decoder_tokenizer.convert_tokens_to_ids('<SEP>')
+        # if pad token ecist then use pad token, othrwise bos token
+        if self.decoder_tokenizer.pad_token_id is None:
+            self.decoder_tokenizer.pad_token_id = self.decoder_tokenizer.bos_token_id
+
+        # resize the tokenizer embedding
+        self.decoder.resize_token_embeddings(len(self.decoder_tokenizer))
+        self.decoder.generation_config.top_p=None
+        self.decoder.generation_config.temperature=None
+        self.compr_model_name = cfg.compr_model_name
+        # other settings
+        self.generation_top_k = cfg.generation_top_k
+        self.sep = cfg.sep
+        self.compr_rate = cfg.compr_rate
+        self.local_rank = os.getenv('LOCAL_RANK', '0')
+
+    def compress_and_replace_emb(self, enc_input_ids, enc_attention_mask, dec_input_ids):
+        indices = range(0, enc_input_ids.size(0) + 1, self.generation_top_k)
+        if self.compr:
+            compressed_embs = self.compr(enc_input_ids, enc_attention_mask)
+            input_embeds = self.replace_embeddings(compressed_embs, dec_input_ids, indices)
+        else:
+            compressed_embs = self.compr_decoder(enc_input_ids, enc_attention_mask)
+            input_embeds = self.replace_embeddings(compressed_embs, dec_input_ids, indices)
+        return input_embeds
+    
+    def compr_decoder(self, input_ids, attention_mask):
+        emb = self.decoder(input_ids=input_ids, attention_mask=attention_mask, output_hidden_states=True).hidden_states[-1]
+        mask = input_ids == self.decoder_tokenizer.mem_token_id
+        return emb[mask].reshape(emb.size(0), -1, emb.size(-1))
+    
+
+    def replace_embeddings(self, compressed_embs, dec_input_ids, indices):
+        # Embed the decoder input
+        inputs_embeds = self.decoder.get_input_embeddings()(dec_input_ids)
+        num_embs = compressed_embs.size(1)
+        if self.sep:
+            slot_len = num_embs + 1
+        else:
+            slot_len = num_embs
+        # get first mem_token inidices
+        first_mem_token_indices = torch.argmax((dec_input_ids == self.decoder_tokenizer.mem_token_id).int(), dim=1)
+        batch_size = inputs_embeds.size(0)
+        # for each example in batch, replace them with compressed embeddings 
+        for i in range(batch_size):
+            for j in range(indices[i], indices[i + 1]):
+                start_idx = first_mem_token_indices[i].item() + (j-indices[i]) * slot_len
+                inputs_embeds[i, start_idx:start_idx + num_embs, :] = compressed_embs[j]
+        return inputs_embeds
+
+
+    def forward(self, 
+            enc_input_ids: torch.LongTensor = None,
+            enc_attention_mask: torch.LongTensor = None,
+            dec_input_ids: torch.LongTensor = None, 
+            dec_attention_mask: torch.LongTensor = None,
+            labels: torch.LongTensor = None):
         
-    def _replace_attention_with_linformer(self):
-        # Replace all attention layers with LinformerSelfAttention in the model
-        for layer in self.decoder.transformer.h:
-            layer.attn = LinformerSelfAttention(
-                dim=layer.attn.attn.in_proj_weight.shape[0],
-                num_heads=layer.attn.num_attention_heads,
-                dropout=0.1,
-                n_heads=layer.attn.num_attention_heads,
-                d_head=layer.attn.attn.in_proj_weight.shape[0] // layer.attn.num_attention_heads
-            )
+        # enc_input_ids: stores the contexts, should be flattened from all queries before input, dimention (batch_size*generation_top_k, token_length)
+        # enc_attention_mask: attention mask of enc_input_ids
+        # dec_input_ids: stores the prompts (including mem tokens), dimention (batch_size, token_length)
+        # dec_attention_mask: attention mask of dec_input_ids
 
-    def _apply_lora(self, lora_r):
-        # Apply LoRA as per your configuration
-        peft_config = LoraConfig(
-            task_type="CAUSAL_LM",
-            r=lora_r,
-            lora_alpha=2 * lora_r,
-            target_modules='all-linear',
-            lora_dropout=0.1,
-        )
-        self.decoder = get_peft_model(self.decoder, peft_config)
-
-    def forward(self, enc_input_ids, enc_attention_mask, dec_input_ids, dec_attention_mask, labels):
+        # Perform compression with gradient tracking
         inputs_embeds = self.compress_and_replace_emb(enc_input_ids, enc_attention_mask, dec_input_ids)
+
+        # if training_form is compressor, then detach the inputs_embeds, to make gradient not count in decoder
+        if (self.training_form == "compressor") and (self.compr is None):
+            inputs_embeds  = inputs_embeds.detach()
+
+        # decoding
         decoder_outputs = self.decoder(inputs_embeds=inputs_embeds, attention_mask=dec_attention_mask, labels=labels)
+
         return {"loss": decoder_outputs.loss, "logits": decoder_outputs.logits}
 
+
+        
     def generate(self, model_input, max_new_tokens=128):
         device = self.decoder.device
         enc_input_ids, enc_attention_mask, dec_input_ids, dec_attention_mask = model_input['enc_input_ids'], model_input['enc_attention_mask'], model_input['dec_input_ids'], model_input['dec_attention_mask']
@@ -149,8 +263,52 @@ class COCOM(PreTrainedModel):
             attention_mask=dec_attention_mask.to(device),
             do_sample=False,
             top_p=None,
-            max_new_tokens=min(max_new_tokens, 4096)
-        )
+            max_new_tokens=max_new_tokens
+            )
         decoded = self.decoder_tokenizer.batch_decode(output_ids, skip_special_tokens=True)
         return decoded
+        
+    def generate_from_text(self, contexts, questions, max_new_tokens=128):
+        # for each question in list give input a list of contexts of equal length 
+        # first make sure that every list in contexts are having the same length
+        assert len(contexts) == len(questions)
+        assert all([len(context) == len(contexts[0]) for context in contexts])
+
+        # prepare inp_enc for compression
+        # first flatten the contexts
+        self.generation_top_k = len(contexts[0])
+        flat_contexts = sum(contexts, [])
+        #tokenize the contexts, depending if compr exist or not
+        if self.compr is not None:
+            enc_input = self.compr.tokenizer(flat_contexts, padding=True, truncation=True, return_tensors='pt', pad_to_multiple_of=self.compr_rate)
+            num_mem_tokens = math.ceil(enc_input['input_ids'].size(1) / self.compr_rate)
+        else:
+            # first need to add special token in flat_contexts
+            flat_contexts = [self.decoder_tokenizer.enc_token + self.decoder_tokenizer.bos_token +  context  + self.decoder_tokenizer.bos_token  for context in flat_contexts]
+            enc_input  = self.decoder_tokenizer(flat_contexts, truncation=True, return_tensors='pt', padding="longest")
+            num_mem_tokens = math.ceil((enc_input['input_ids'].size(1)-3) / self.compr_rate)
+            mem_tokens = torch.full((enc_input['input_ids'].size(0), num_mem_tokens), self.decoder_tokenizer.mem_token_id, dtype=torch.long)
+            enc_input['input_ids'] = torch.cat([mem_tokens, enc_input['input_ids']], dim=1)
+            enc_input['attention_mask'] = torch.cat([torch.ones_like(mem_tokens), enc_input['attention_mask']], dim=1)
+        
+        
+        # prepare inp_dec
+        mem_tokens = self.decoder_tokenizer.mem_token * num_mem_tokens
+        if self.sep:
+            mem_tokens += self.decoder_tokenizer.sep_token
+        
+        instr = [self.decoder_tokenizer.bos_token + mem_tokens* self.generation_top_k + '[INST]' + question + '\n[/INST]\n' for question in questions]
+        inp_dec = self.decoder_tokenizer(instr, truncation=True, return_tensors='pt', padding="longest")
+
+        # generate
+        model_input = {
+            'enc_input_ids': enc_input['input_ids'].to(self.decoder.device),
+            'enc_attention_mask': enc_input['attention_mask'].to(self.decoder.device),
+            'dec_input_ids': inp_dec['input_ids'].to(self.decoder.device),
+            'dec_attention_mask': inp_dec['attention_mask'].to(self.decoder.device)
+        }
+
+        return self.generate(model_input, max_new_tokens)
+        
 
+