Upload 28 files

.gitattributes

@@ -33,3 +33,7 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+4BIT_FP/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+4BIT_NF/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+4BIT/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+8BIT/tokenizer.json filter=lfs diff=lfs merge=lfs -text

4BIT/4BIT.py

@@ -0,0 +1,18 @@
+from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig
+import torch
+
+### Config Quantização
+quantization = BitsAndBytesConfig (
+    load_in_4bit = True,
+)
+
+# Caminho do modelo caso seja self hosted
+PATH = r""
+# Nome do modelo HF
+MODEL = "amalia-llmAMALIA-9B-0626-SFT"
+
+modelo = AutoModelForCausalLM.from_pretrained (PATH, device_map = "auto", quantization_config = quantization)
+tokenizer = AutoTokenizer.from_pretrained (PATH)
+
+modelo.save_pretrained (r"")
+tokenizer.save_pretrained (r"")

4BIT/chat_template.jinja

@@ -0,0 +1,4 @@
+{% if messages[0]['role'] != 'system' %}{% set messages = [{'role': 'system', 'content': 'O teu nome é AMALIA, e és um modelo avançado de linguagem útil. Responde sempre na língua do utilizador, a menos que sejas instruído em contrário, e lembra-te que a tua língua principal é o português europeu.'}] + messages %}{% endif %}{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% for message in messages %}{{ '<|im_start|>' + message['role'] + '
+' + message['content'] + '<|im_end|>
+' }}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
+' }}{% endif %}

4BIT/config.json

@@ -0,0 +1,47 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "dtype": "bfloat16",
+  "eos_token_id": 4,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 12288,
+  "max_position_embeddings": 32768,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 2,
+  "pretraining_tp": 1,
+  "quantization_config": {
+    "_load_in_4bit": true,
+    "_load_in_8bit": false,
+    "bnb_4bit_compute_dtype": "float32",
+    "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": null,
+    "llm_int8_threshold": 6.0,
+    "load_in_4bit": true,
+    "load_in_8bit": false,
+    "quant_method": "bitsandbytes"
+  },
+  "rms_norm_eps": 1e-05,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": false,
+  "transformers_version": "5.12.1",
+  "use_cache": false,
+  "vocab_size": 128000
+}

4BIT/generation_config.json

@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "do_sample": true,
+  "eos_token_id": [
+    4,
+    2
+  ],
+  "pad_token_id": 2,
+  "transformers_version": "5.12.1"
+}

4BIT/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9dd7f6e08d31ca4d490e4682dcc1670331ecd194dc519b568ad5f4b2f60af770
+size 6656203650

4BIT/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40b8655634176c8f420b2eddd01ce01081902c54e56fd604c19a1ea4d9fbe331
+size 15782929

4BIT/tokenizer_config.json

@@ -0,0 +1,16 @@
+{
+  "add_prefix_space": true,
+  "backend": "tokenizers",
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "is_local": true,
+  "local_files_only": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "</s>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

4BIT_FP/4BITFP.py

@@ -0,0 +1,22 @@
+from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig
+import torch
+
+### Config Quantização
+quantization = BitsAndBytesConfig (
+    load_in_4bit = True,
+    bnb_4bit_quant_type = 'fp4',  #Tipo de Quantização
+    bnb_4bit_use_double_quant = True, #Double Quantization
+    bnb_4bit_compute_dtype = 'float16', #Tipo de precisão usada nos cálculos durante a inferência
+)
+
+
+# Caminho do modelo caso seja self hosted
+PATH = r""
+# Nome do modelo HF
+MODEL = "amalia-llmAMALIA-9B-0626-SFT"
+
+modelo = AutoModelForCausalLM.from_pretrained (PATH, device_map = "auto", quantization_config = quantization)
+tokenizer = AutoTokenizer.from_pretrained (PATH)
+
+modelo.save_pretrained (r"")
+tokenizer.save_pretrained (r"")

4BIT_FP/chat_template.jinja

@@ -0,0 +1,4 @@
+{% if messages[0]['role'] != 'system' %}{% set messages = [{'role': 'system', 'content': 'O teu nome é AMALIA, e és um modelo avançado de linguagem útil. Responde sempre na língua do utilizador, a menos que sejas instruído em contrário, e lembra-te que a tua língua principal é o português europeu.'}] + messages %}{% endif %}{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% for message in messages %}{{ '<|im_start|>' + message['role'] + '
+' + message['content'] + '<|im_end|>
+' }}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
+' }}{% endif %}

4BIT_FP/config.json

@@ -0,0 +1,47 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "dtype": "bfloat16",
+  "eos_token_id": 4,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 12288,
+  "max_position_embeddings": 32768,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 2,
+  "pretraining_tp": 1,
+  "quantization_config": {
+    "_load_in_4bit": true,
+    "_load_in_8bit": false,
+    "bnb_4bit_compute_dtype": "float16",
+    "bnb_4bit_quant_storage": "uint8",
+    "bnb_4bit_quant_type": "fp4",
+    "bnb_4bit_use_double_quant": true,
+    "llm_int8_enable_fp32_cpu_offload": false,
+    "llm_int8_has_fp16_weight": false,
+    "llm_int8_skip_modules": null,
+    "llm_int8_threshold": 6.0,
+    "load_in_4bit": true,
+    "load_in_8bit": false,
+    "quant_method": "bitsandbytes"
+  },
+  "rms_norm_eps": 1e-05,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": false,
+  "transformers_version": "5.12.1",
+  "use_cache": false,
+  "vocab_size": 128000
+}

4BIT_FP/generation_config.json

@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "do_sample": true,
+  "eos_token_id": [
+    4,
+    2
+  ],
+  "pad_token_id": 2,
+  "transformers_version": "5.12.1"
+}

4BIT_FP/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12a503101f8a5aa8af2440f31382e6f9406a31a0b7bd9786600e4c8e4e180212
+size 6278731678

4BIT_FP/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40b8655634176c8f420b2eddd01ce01081902c54e56fd604c19a1ea4d9fbe331
+size 15782929

4BIT_FP/tokenizer_config.json

@@ -0,0 +1,16 @@
+{
+  "add_prefix_space": true,
+  "backend": "tokenizers",
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "is_local": true,
+  "local_files_only": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "</s>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

4BIT_NF/4BITNF.py

@@ -0,0 +1,22 @@
+from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig
+import torch
+
+### Config Quantização
+quantization = BitsAndBytesConfig (
+    load_in_4bit = True,
+    bnb_4bit_quant_type = 'nf4',  #Tipo de Quantização
+    bnb_4bit_use_double_quant = True, #Double Quantization
+    bnb_4bit_compute_dtype = 'float16', #Tipo de precisão usada nos cálculos durante a inferência
+)
+
+
+# Caminho do modelo caso seja self hosted
+PATH = r""
+# Nome do modelo HF
+MODEL = "amalia-llmAMALIA-9B-0626-SFT"
+
+modelo = AutoModelForCausalLM.from_pretrained (PATH, device_map = "auto", quantization_config = quantization)
+tokenizer = AutoTokenizer.from_pretrained (PATH)
+
+modelo.save_pretrained (r"")
+tokenizer.save_pretrained (r"")

4BIT_NF/chat_template.jinja

@@ -0,0 +1,4 @@
+{% if messages[0]['role'] != 'system' %}{% set messages = [{'role': 'system', 'content': 'O teu nome é AMALIA, e és um modelo avançado de linguagem útil. Responde sempre na língua do utilizador, a menos que sejas instruído em contrário, e lembra-te que a tua língua principal é o português europeu.'}] + messages %}{% endif %}{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% for message in messages %}{{ '<|im_start|>' + message['role'] + '
+' + message['content'] + '<|im_end|>
+' }}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
+' }}{% endif %}

4BIT_NF/config.json

@@ -0,0 +1,47 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "dtype": "bfloat16",
+  "eos_token_id": 4,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 12288,
+  "max_position_embeddings": 32768,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 2,
+  "pretraining_tp": 1,
+  "quantization_config": {
+    "_load_in_4bit": true,
+    "_load_in_8bit": false,
+    "bnb_4bit_compute_dtype": "float16",
+    "bnb_4bit_quant_storage": "uint8",
+    "bnb_4bit_quant_type": "nf4",
+    "bnb_4bit_use_double_quant": true,
+    "llm_int8_enable_fp32_cpu_offload": false,
+    "llm_int8_has_fp16_weight": false,
+    "llm_int8_skip_modules": null,
+    "llm_int8_threshold": 6.0,
+    "load_in_4bit": true,
+    "load_in_8bit": false,
+    "quant_method": "bitsandbytes"
+  },
+  "rms_norm_eps": 1e-05,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": false,
+  "transformers_version": "5.12.1",
+  "use_cache": false,
+  "vocab_size": 128000
+}

4BIT_NF/generation_config.json

@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "do_sample": true,
+  "eos_token_id": [
+    4,
+    2
+  ],
+  "pad_token_id": 2,
+  "transformers_version": "5.12.1"
+}

4BIT_NF/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a3b063c1ef7bb8c2b4102dc8e991deee5c962d498db61a66fdcef41659fc927
+size 6278731678

4BIT_NF/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40b8655634176c8f420b2eddd01ce01081902c54e56fd604c19a1ea4d9fbe331
+size 15782929

4BIT_NF/tokenizer_config.json

@@ -0,0 +1,16 @@
+{
+  "add_prefix_space": true,
+  "backend": "tokenizers",
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "is_local": true,
+  "local_files_only": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "</s>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

8BIT/8BIT.py

@@ -0,0 +1,18 @@
+from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig
+import torch
+
+### Config Quantização
+quantization = BitsAndBytesConfig (
+    load_in_8bit = True,
+)
+
+# Caminho do modelo caso seja self hosted
+PATH = r""
+# Nome do modelo HF
+MODEL = "amalia-llmAMALIA-9B-0626-SFT"
+
+modelo = AutoModelForCausalLM.from_pretrained (PATH, device_map = "auto", quantization_config = quantization)
+tokenizer = AutoTokenizer.from_pretrained (PATH)
+
+modelo.save_pretrained (r"")
+tokenizer.save_pretrained (r"")

8BIT/chat_template.jinja

@@ -0,0 +1,4 @@
+{% if messages[0]['role'] != 'system' %}{% set messages = [{'role': 'system', 'content': 'O teu nome é AMALIA, e és um modelo avançado de linguagem útil. Responde sempre na língua do utilizador, a menos que sejas instruído em contrário, e lembra-te que a tua língua principal é o português europeu.'}] + messages %}{% endif %}{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% for message in messages %}{{ '<|im_start|>' + message['role'] + '
+' + message['content'] + '<|im_end|>
+' }}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
+' }}{% endif %}

8BIT/config.json

@@ -0,0 +1,47 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "dtype": "bfloat16",
+  "eos_token_id": 4,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 12288,
+  "max_position_embeddings": 32768,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 2,
+  "pretraining_tp": 1,
+  "quantization_config": {
+    "_load_in_4bit": false,
+    "_load_in_8bit": true,
+    "bnb_4bit_compute_dtype": "float32",
+    "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": null,
+    "llm_int8_threshold": 6.0,
+    "load_in_4bit": false,
+    "load_in_8bit": true,
+    "quant_method": "bitsandbytes"
+  },
+  "rms_norm_eps": 1e-05,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "tie_word_embeddings": false,
+  "transformers_version": "5.12.1",
+  "use_cache": false,
+  "vocab_size": 128000
+}

8BIT/generation_config.json

@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "do_sample": true,
+  "eos_token_id": [
+    4,
+    2
+  ],
+  "pad_token_id": 2,
+  "transformers_version": "5.12.1"
+}

8BIT/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5c202015799a2c08fabc44c612f59082eb765b05b36c6a889a043ef2431103c0
+size 10207888278

8BIT/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40b8655634176c8f420b2eddd01ce01081902c54e56fd604c19a1ea4d9fbe331
+size 15782929

8BIT/tokenizer_config.json

@@ -0,0 +1,16 @@
+{
+  "add_prefix_space": true,
+  "backend": "tokenizers",
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "is_local": true,
+  "local_files_only": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "</s>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}