Views
No views yet
| Name | Quant method | Size |
|---|---|---|
| Neurona-2b.Q2_K.gguf | Q2_K | 1.08GB |
| Neurona-2b.IQ3_XS.gguf | IQ3_XS | 1.16GB |
| Neurona-2b.IQ3_S.gguf | IQ3_S | 1.2GB |
| Neurona-2b.Q3_K_S.gguf | Q3_K_S | 1.2GB |
| Neurona-2b.IQ3_M.gguf | IQ3_M | 1.22GB |
| Neurona-2b.Q3_K.gguf | Q3_K | 1.29GB |
| Neurona-2b.Q3_K_M.gguf | Q3_K_M | 1.29GB |
| Neurona-2b.Q3_K_L.gguf | Q3_K_L | 1.36GB |
| Neurona-2b.IQ4_XS.gguf | IQ4_XS | 1.4GB |
| Neurona-2b.Q4_0.gguf | Q4_0 | 1.44GB |
| Neurona-2b.IQ4_NL.gguf | IQ4_NL | 1.45GB |
| Neurona-2b.Q4_K_S.gguf | Q4_K_S | 1.45GB |
| Neurona-2b.Q4_K.gguf | Q4_K | 1.52GB |
| Neurona-2b.Q4_K_M.gguf | Q4_K_M | 1.52GB |
| Neurona-2b.Q4_1.gguf | Q4_1 | 1.56GB |
| Neurona-2b.Q5_0.gguf | Q5_0 | 1.68GB |
| Neurona-2b.Q5_K_S.gguf | Q5_K_S | 1.68GB |
| Neurona-2b.Q5_K.gguf | Q5_K | 1.71GB |
| Neurona-2b.Q5_K_M.gguf | Q5_K_M | 1.71GB |
| Neurona-2b.Q5_1.gguf | Q5_1 | 1.79GB |
| Neurona-2b.Q6_K.gguf | Q6_K | 1.92GB |
| Neurona-2b.Q8_0.gguf | Q8_0 | 2.49GB |
Esta es una versión preliminar del dataset card. El modelo está en desarrollo y no es la versión final. Si quieres saber más sobre este modelo, escribe a iker.garciaf@ehu.eus
1base_model: google/gemma-2b
2model_type: AutoModelForCausalLM
3tokenizer_type: AutoTokenizer
4is_falcon_derived_model:
5is_llama_derived_model:
6is_qwen_derived_model:
7is_mistral_derived_model:
8
9load_in_8bit: false
10load_in_4bit: false
11strict: false
12
13device_map: null
14
15datasets:
16 - path: /ikerlariak/igarcia945/Mortadelo-Filemon/final_dataset/OpenHermes-2.5-Spanish_fix_gpt.jsonl
17 type: sharegpt
18 conversation: chatml
19 field: conversations
20 roles:
21 input:
22 - system
23 - gpt
24 output:
25 - human
26 - path: /ikerlariak/igarcia945/Mortadelo-Filemon/final_dataset/OpenHermes-2.5-English.jsonl
27 type: sharegpt
28 conversation: chatml
29 field: conversations
30 - path: /ikerlariak/igarcia945/Mortadelo-Filemon/final_dataset/glaive-function-calling-v2.jsonl
31 type: sharegpt
32 conversation: chatml
33 field: conversations
34 roles:
35 input:
36 - system
37 - gpt
38 - tool
39 output:
40 - human
41 - path: /ikerlariak/igarcia945/Mortadelo-Filemon/final_dataset/glaive-code-assistant-v3-small.jsonl
42 type: sharegpt
43 conversation: chatml
44 field: conversations
45 roles:
46 input:
47 - system
48 - gpt
49 output:
50 - human
51chat_template: chatml
52
53dataset_prepared_path: /ikerlariak/igarcia945/Mortadelo-Filemon/gemma-2b-spanish/dataset
54
55shuffle_merged_datasets: true
56
57val_set_size: 0.005
58
59output_dir: /ikerlariak/igarcia945/Mortadelo-Filemon/gemma-2b-spanish/
60
61adapter:
62lora_model_dir:
63
64sequence_len: 8192
65sample_packing: true
66eval_sample_packing: false
67pad_to_sequence_len: false
68
69special_tokens:
70 bos_token: "<|im_start|>"
71 eos_token: "<|im_end|>"
72 pad_token: "<|end_of_text|>"
73
74tokens:
75 - "<|begin_of_text|>"
76 - "<|end_of_text|>"
77 - "<|im_start|>"
78 - "<|im_end|>"
79 - "<|start_header_id|>"
80 - "<|end_header_id|>"
81 - "<tool_call>"
82 - "<tool_response>"
83 - "<tools>"
84 - "</tool_call>"
85 - "</tool_response>"
86 - "</tools>"
87 - "<reserved1>"
88 - "<reserved2>"
89 - "<reserved3>"
90 - "<reserved4>"
91
92
93
94neftune_noise_alpha: 5
95
96wandb_project: Mortadelo&Filemon
97wandb_entity: igarciaf
98wandb_watch:
99wandb_name: gemma2b
100wandb_log_model:
101
102gradient_accumulation_steps: 32
103micro_batch_size: 2
104eval_batch_size: 2
105num_epochs: 3
106optimizer: adamw_torch_fused
107lr_scheduler: cosine
108learning_rate: 0.00007
109
110
111train_on_inputs: false
112group_by_length: false
113bf16: true
114fp16: false
115tf32: false
116
117gradient_checkpointing: true
118early_stopping_patience:
119resume_from_checkpoint:
120local_rank:
121logging_steps: 1
122xformers_attention:
123flash_attention: true
124
125warmup_ratio: 0.03
126evals_per_epoch: 4
127eval_table_size:
128save_strategy: "no"
129debug:
130deepspeed: /ikerlariak/igarcia945/Mortadelo-Filemon/train_configs/deepspeed_zero3.json
131weight_decay: 0.0
132fsdp:
133fsdp_config:
134special_tokens:
135
136seed: 33