Update README.md
Browse files
README.md
CHANGED
|
@@ -13,4 +13,168 @@ tags:
|
|
| 13 |
|
| 14 |
# EXAONE-3.0-8B-it
|
| 15 |
|
| 16 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
|
| 14 |
# EXAONE-3.0-8B-it
|
| 15 |
|
| 16 |
+
```py
|
| 17 |
+
from llama_cpp import Llama
|
| 18 |
+
|
| 19 |
+
llm = Llama.from_pretrained(
|
| 20 |
+
repo_id="Bingsu/exaone-3.0-7.8b-it",
|
| 21 |
+
filename="exaone-3.0-7.8B-it-Q8_0.gguf"
|
| 22 |
+
)
|
| 23 |
+
```
|
| 24 |
+
|
| 25 |
+
```sh
|
| 26 |
+
llama_model_loader: loaded meta data with 34 key-value pairs and 291 tensors from /root/.cache/huggingface/hub/models--Bingsu--exaone-3.0-7.8b-it/snapshots/c7b9c43a7d1db6509b40e9b18f10ae0554b3d4cb/./exaone-3.0-7.8B-it-Q8_0.gguf (version GGUF V3 (latest))
|
| 27 |
+
llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
|
| 28 |
+
llama_model_loader: - kv 0: general.architecture str = llama
|
| 29 |
+
llama_model_loader: - kv 1: general.type str = model
|
| 30 |
+
llama_model_loader: - kv 2: general.name str = Exaone 3.0 7.8b It
|
| 31 |
+
llama_model_loader: - kv 3: general.finetune str = it
|
| 32 |
+
llama_model_loader: - kv 4: general.basename str = exaone-3.0
|
| 33 |
+
llama_model_loader: - kv 5: general.size_label str = 7.8B
|
| 34 |
+
llama_model_loader: - kv 6: general.license str = other
|
| 35 |
+
llama_model_loader: - kv 7: general.license.name str = exaone
|
| 36 |
+
llama_model_loader: - kv 8: general.license.link str = LICENSE
|
| 37 |
+
llama_model_loader: - kv 9: general.tags arr[str,2] = ["lg-ai", "exaone"]
|
| 38 |
+
llama_model_loader: - kv 10: general.languages arr[str,2] = ["en", "ko"]
|
| 39 |
+
llama_model_loader: - kv 11: llama.block_count u32 = 32
|
| 40 |
+
llama_model_loader: - kv 12: llama.context_length u32 = 4096
|
| 41 |
+
llama_model_loader: - kv 13: llama.embedding_length u32 = 4096
|
| 42 |
+
llama_model_loader: - kv 14: llama.feed_forward_length u32 = 14336
|
| 43 |
+
llama_model_loader: - kv 15: llama.attention.head_count u32 = 32
|
| 44 |
+
llama_model_loader: - kv 16: llama.attention.head_count_kv u32 = 8
|
| 45 |
+
llama_model_loader: - kv 17: llama.rope.freq_base f32 = 500000.000000
|
| 46 |
+
llama_model_loader: - kv 18: llama.attention.layer_norm_rms_epsilon f32 = 0.000010
|
| 47 |
+
llama_model_loader: - kv 19: general.file_type u32 = 7
|
| 48 |
+
llama_model_loader: - kv 20: llama.vocab_size u32 = 102400
|
| 49 |
+
llama_model_loader: - kv 21: llama.rope.dimension_count u32 = 128
|
| 50 |
+
llama_model_loader: - kv 22: tokenizer.ggml.add_space_prefix bool = false
|
| 51 |
+
llama_model_loader: - kv 23: tokenizer.ggml.model str = gpt2
|
| 52 |
+
llama_model_loader: - kv 24: tokenizer.ggml.pre str = default
|
| 53 |
+
llama_model_loader: - kv 25: tokenizer.ggml.tokens arr[str,102400] = ["[PAD]", "[BOS]", "[EOS]", "[UNK]", ...
|
| 54 |
+
llama_model_loader: - kv 26: tokenizer.ggml.token_type arr[i32,102400] = [3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4, ...
|
| 55 |
+
llama_model_loader: - kv 27: tokenizer.ggml.merges arr[str,101782] = ["t h", "Δ a", "Δ Γ", "i n", "Δ t...
|
| 56 |
+
llama_model_loader: - kv 28: tokenizer.ggml.bos_token_id u32 = 1
|
| 57 |
+
llama_model_loader: - kv 29: tokenizer.ggml.eos_token_id u32 = 361
|
| 58 |
+
llama_model_loader: - kv 30: tokenizer.ggml.unknown_token_id u32 = 3
|
| 59 |
+
llama_model_loader: - kv 31: tokenizer.ggml.padding_token_id u32 = 0
|
| 60 |
+
llama_model_loader: - kv 32: tokenizer.chat_template str = {% for message in messages %}{% if lo...
|
| 61 |
+
llama_model_loader: - kv 33: general.quantization_version u32 = 2
|
| 62 |
+
llama_model_loader: - type f32: 65 tensors
|
| 63 |
+
llama_model_loader: - type q8_0: 226 tensors
|
| 64 |
+
llm_load_vocab: special tokens cache size = 362
|
| 65 |
+
llm_load_vocab: token to piece cache size = 0.6622 MB
|
| 66 |
+
llm_load_print_meta: format = GGUF V3 (latest)
|
| 67 |
+
llm_load_print_meta: arch = llama
|
| 68 |
+
llm_load_print_meta: vocab type = BPE
|
| 69 |
+
llm_load_print_meta: n_vocab = 102400
|
| 70 |
+
llm_load_print_meta: n_merges = 101782
|
| 71 |
+
llm_load_print_meta: vocab_only = 0
|
| 72 |
+
llm_load_print_meta: n_ctx_train = 4096
|
| 73 |
+
llm_load_print_meta: n_embd = 4096
|
| 74 |
+
llm_load_print_meta: n_layer = 32
|
| 75 |
+
llm_load_print_meta: n_head = 32
|
| 76 |
+
llm_load_print_meta: n_head_kv = 8
|
| 77 |
+
llm_load_print_meta: n_rot = 128
|
| 78 |
+
llm_load_print_meta: n_swa = 0
|
| 79 |
+
llm_load_print_meta: n_embd_head_k = 128
|
| 80 |
+
llm_load_print_meta: n_embd_head_v = 128
|
| 81 |
+
llm_load_print_meta: n_gqa = 4
|
| 82 |
+
llm_load_print_meta: n_embd_k_gqa = 1024
|
| 83 |
+
llm_load_print_meta: n_embd_v_gqa = 1024
|
| 84 |
+
llm_load_print_meta: f_norm_eps = 0.0e+00
|
| 85 |
+
llm_load_print_meta: f_norm_rms_eps = 1.0e-05
|
| 86 |
+
llm_load_print_meta: f_clamp_kqv = 0.0e+00
|
| 87 |
+
llm_load_print_meta: f_max_alibi_bias = 0.0e+00
|
| 88 |
+
llm_load_print_meta: f_logit_scale = 0.0e+00
|
| 89 |
+
llm_load_print_meta: n_ff = 14336
|
| 90 |
+
llm_load_print_meta: n_expert = 0
|
| 91 |
+
llm_load_print_meta: n_expert_used = 0
|
| 92 |
+
llm_load_print_meta: causal attn = 1
|
| 93 |
+
llm_load_print_meta: pooling type = 0
|
| 94 |
+
llm_load_print_meta: rope type = 0
|
| 95 |
+
llm_load_print_meta: rope scaling = linear
|
| 96 |
+
llm_load_print_meta: freq_base_train = 500000.0
|
| 97 |
+
llm_load_print_meta: freq_scale_train = 1
|
| 98 |
+
llm_load_print_meta: n_ctx_orig_yarn = 4096
|
| 99 |
+
llm_load_print_meta: rope_finetuned = unknown
|
| 100 |
+
llm_load_print_meta: ssm_d_conv = 0
|
| 101 |
+
llm_load_print_meta: ssm_d_inner = 0
|
| 102 |
+
llm_load_print_meta: ssm_d_state = 0
|
| 103 |
+
llm_load_print_meta: ssm_dt_rank = 0
|
| 104 |
+
llm_load_print_meta: model type = 8B
|
| 105 |
+
llm_load_print_meta: model ftype = Q8_0
|
| 106 |
+
llm_load_print_meta: model params = 7.82 B
|
| 107 |
+
llm_load_print_meta: model size = 7.74 GiB (8.50 BPW)
|
| 108 |
+
llm_load_print_meta: general.name = Exaone 3.0 7.8b It
|
| 109 |
+
llm_load_print_meta: BOS token = 1 '[BOS]'
|
| 110 |
+
llm_load_print_meta: EOS token = 361 '[|endofturn|]'
|
| 111 |
+
llm_load_print_meta: UNK token = 3 '[UNK]'
|
| 112 |
+
llm_load_print_meta: PAD token = 0 '[PAD]'
|
| 113 |
+
llm_load_print_meta: LF token = 490 'Γ'
|
| 114 |
+
llm_load_print_meta: EOT token = 42 '<|endoftext|>'
|
| 115 |
+
llm_load_print_meta: max token length = 48
|
| 116 |
+
ggml_cuda_init: GGML_CUDA_FORCE_MMQ: yes
|
| 117 |
+
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
|
| 118 |
+
ggml_cuda_init: found 1 CUDA devices:
|
| 119 |
+
Device 0: NVIDIA L4, compute capability 8.9, VMM: yes
|
| 120 |
+
llm_load_tensors: ggml ctx size = 0.14 MiB
|
| 121 |
+
llm_load_tensors: offloading 0 repeating layers to GPU
|
| 122 |
+
llm_load_tensors: offloaded 0/33 layers to GPU
|
| 123 |
+
llm_load_tensors: CPU buffer size = 7923.02 MiB
|
| 124 |
+
............................................................................................
|
| 125 |
+
llama_new_context_with_model: n_ctx = 512
|
| 126 |
+
llama_new_context_with_model: n_batch = 512
|
| 127 |
+
llama_new_context_with_model: n_ubatch = 512
|
| 128 |
+
llama_new_context_with_model: flash_attn = 0
|
| 129 |
+
llama_new_context_with_model: freq_base = 500000.0
|
| 130 |
+
llama_new_context_with_model: freq_scale = 1
|
| 131 |
+
llama_kv_cache_init: CUDA_Host KV buffer size = 64.00 MiB
|
| 132 |
+
llama_new_context_with_model: KV self size = 64.00 MiB, K (f16): 32.00 MiB, V (f16): 32.00 MiB
|
| 133 |
+
llama_new_context_with_model: CUDA_Host output buffer size = 0.39 MiB
|
| 134 |
+
llama_new_context_with_model: CUDA0 compute buffer size = 633.00 MiB
|
| 135 |
+
llama_new_context_with_model: CUDA_Host compute buffer size = 9.01 MiB
|
| 136 |
+
llama_new_context_with_model: graph nodes = 1030
|
| 137 |
+
llama_new_context_with_model: graph splits = 356
|
| 138 |
+
AVX = 1 | AVX_VNNI = 0 | AVX2 = 1 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | FMA = 1 | NEON = 0 | SVE = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 |
|
| 139 |
+
Model metadata: {'tokenizer.ggml.unknown_token_id': '3', 'tokenizer.ggml.eos_token_id': '361', 'general.quantization_version': '2', 'tokenizer.ggml.model': 'gpt2', 'tokenizer.ggml.add_space_prefix': 'false', 'llama.rope.dimension_count': '128', 'llama.vocab_size': '102400', 'general.file_type': '7', 'llama.attention.layer_norm_rms_epsilon': '0.000010', 'llama.rope.freq_base': '500000.000000', 'tokenizer.ggml.bos_token_id': '1', 'llama.attention.head_count': '32', 'general.architecture': 'llama', 'llama.attention.head_count_kv': '8', 'llama.block_count': '32', 'tokenizer.ggml.padding_token_id': '0', 'general.basename': 'exaone-3.0', 'tokenizer.ggml.pre': 'default', 'llama.context_length': '4096', 'general.name': 'Exaone 3.0 7.8b It', 'general.type': 'model', 'general.size_label': '7.8B', 'general.finetune': 'it', 'general.license.name': 'exaone', 'tokenizer.chat_template': "{% for message in messages %}{% if loop.first and message['role'] != 'system' %}{{ '[|system|][|endofturn|]\n' }}{% endif %}{{ '[|' + message['role'] + '|]' + message['content'] }}{% if message['role'] == 'user' %}{{ '\n' }}{% else %}{{ '[|endofturn|]\n' }}{% endif %}{% endfor %}{% if add_generation_prompt %}{{ '[|assistant|]' }}{% endif %}", 'general.license.link': 'LICENSE', 'general.license': 'other', 'llama.feed_forward_length': '14336', 'llama.embedding_length': '4096'}
|
| 140 |
+
Available chat formats from metadata: chat_template.default
|
| 141 |
+
Using gguf chat template: {% for message in messages %}{% if loop.first and message['role'] != 'system' %}{{ '[|system|][|endofturn|]
|
| 142 |
+
' }}{% endif %}{{ '[|' + message['role'] + '|]' + message['content'] }}{% if message['role'] == 'user' %}{{ '
|
| 143 |
+
' }}{% else %}{{ '[|endofturn|]
|
| 144 |
+
' }}{% endif %}{% endfor %}{% if add_generation_prompt %}{{ '[|assistant|]' }}{% endif %}
|
| 145 |
+
Using chat eos_token: [|endofturn|]
|
| 146 |
+
Using chat bos_token: [BOS]
|
| 147 |
+
```
|
| 148 |
+
|
| 149 |
+
```py
|
| 150 |
+
llm.create_chat_completion(
|
| 151 |
+
messages = [
|
| 152 |
+
{
|
| 153 |
+
"role": "system",
|
| 154 |
+
"content": "You are EXAONE model from LG AI Research, a helpful assistant."
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"role": "user",
|
| 158 |
+
"content": "λ€ ν΄μ€¬μμ"
|
| 159 |
+
}
|
| 160 |
+
]
|
| 161 |
+
)
|
| 162 |
+
```
|
| 163 |
+
|
| 164 |
+
```sh
|
| 165 |
+
llama_print_timings: load time = 1812.86 ms
|
| 166 |
+
llama_print_timings: sample time = 20.39 ms / 220 runs ( 0.09 ms per token, 10788.54 tokens per second)
|
| 167 |
+
llama_print_timings: prompt eval time = 1812.72 ms / 38 tokens ( 47.70 ms per token, 20.96 tokens per second)
|
| 168 |
+
llama_print_timings: eval time = 33280.46 ms / 219 runs ( 151.97 ms per token, 6.58 tokens per second)
|
| 169 |
+
llama_print_timings: total time = 35397.95 ms / 257 tokens
|
| 170 |
+
{'id': 'chatcmpl-451b0538-c70d-45f4-924b-106f5ac3c02f',
|
| 171 |
+
'object': 'chat.completion',
|
| 172 |
+
'created': 1723204952,
|
| 173 |
+
'model': '/root/.cache/huggingface/hub/models--Bingsu--exaone-3.0-7.8b-it/snapshots/c7b9c43a7d1db6509b40e9b18f10ae0554b3d4cb/./exaone-3.0-7.8B-it-Q8_0.gguf',
|
| 174 |
+
'choices': [{'index': 0,
|
| 175 |
+
'message': {'role': 'assistant',
|
| 176 |
+
'content': 'λ€, μκ² μ΅λλ€. μ΄μ μ λ§μνμ λ΄μ©μ μμ½ν΄ λλ¦¬κ² μ΅λλ€:\n\n1. EXAONE 2.0 λͺ¨λΈμ νΉμ§:\n - 7.8B instruction νλ νλΌλ―Έν°\n - νκ΅μ΄μ μμ΄μμ μ°μν μ±λ₯\n - λ€μν μμ
μμ λμ μ νλ\n\n2. μ°κ΅¬ λ
Όλ¬Έ:\n - "EXAONE 2.0: An Open-Retrieval Large Language Model for Dense Retrieval and Question Answering"\n\n3. μ£Όμ μ±κ³Ό:\n - νκ΅μ΄μ μμ΄μμ μ°μν μ±λ₯\n - λ€μν μμ
μμ λμ μ νλ\n\n4. νμ© μ¬λ‘:\n - κ³ κ° μ§μ μ±λ΄\n - λ²λ₯ λ¬Έμ μμ½\n - μλ£ μ 보 μ 곡\n\n5. κΈ°μ μ μΈλΆ μ¬ν:\n - 7.8B instruction νλ νλΌλ―Έν°\n - νκ΅μ΄μ μμ΄μμ μ°μν μ±λ₯\n - λ€μν μμ
μμ λμ μ νλ\n\nμ΄ μΈμ μΆκ°λ‘ κΆκΈν μ¬νμ΄ μμΌμλ©΄ μΈμ λ μ§ λ§μν΄ μ£ΌμΈμ!'},
|
| 177 |
+
'logprobs': None,
|
| 178 |
+
'finish_reason': 'stop'}],
|
| 179 |
+
'usage': {'prompt_tokens': 38, 'completion_tokens': 219, 'total_tokens': 257}}
|
| 180 |
+
```
|