[Bug]: 文档问答会忽略部分数据，比如证书号是12345 回答的是2345
daimashenjing opened this issue · 7 comments
daimashenjing commented
Model Series
Qwen2.5
What are the models used?
Qwen2.5-7B-Instruct-fp16-gguf
What is the scenario where the problem happened?
ollama
Is this a known issue?
- I have followed the GitHub README.
- I have checked the Qwen documentation and cannot find an answer there.
- I have checked the documentation of the related framework and cannot find useful information.
- I have searched the issues and there is not a similar one.
Information about environment
OS: Ubuntu 22.04
Python: Python 3.11.7
GPUs: 2 * 3090
NVIDIA driver: 535.183.01
CUDA compiler:12.2
PyTorch:none
Log output
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[2113347]: INFO [main] HTTP server listening | hostname="127.0.0.1" n_threads_http="15" port="44671" tid="129702541623296" timestamp=1727600275
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: loaded meta data with 29 key-value pairs and 339 tensors from /home/ollama/.ollama/models/blobs/sha256-efa98abc3476e6f28591527328e4045e6708c0b0bcf4b3ae836c5cc0435976d8 (version GGUF V3 (latest))
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 0: general.architecture str = qwen2
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 1: general.type str = model
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 2: general.name str = qwen2.5-7b-instruct
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 3: general.version str = v0.1
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 4: general.finetune str = qwen2.5-7b-instruct
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 5: general.size_label str = 7.6B
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 6: qwen2.block_count u32 = 28
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 7: qwen2.context_length u32 = 131072
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 8: qwen2.embedding_length u32 = 3584
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 9: qwen2.feed_forward_length u32 = 18944
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 10: qwen2.attention.head_count u32 = 28
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 11: qwen2.attention.head_count_kv u32 = 4
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 12: qwen2.rope.freq_base f32 = 1000000.000000
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 13: qwen2.attention.layer_norm_rms_epsilon f32 = 0.000001
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 14: general.file_type u32 = 1
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 15: tokenizer.ggml.model str = gpt2
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 16: tokenizer.ggml.pre str = qwen2
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 17: tokenizer.ggml.tokens arr[str,152064] = ["!", "\"", "#", "$", "%", "&", "'", ...
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 18: tokenizer.ggml.token_type arr[i32,152064] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 19: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 20: tokenizer.ggml.eos_token_id u32 = 151645
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 21: tokenizer.ggml.padding_token_id u32 = 151643
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 22: tokenizer.ggml.bos_token_id u32 = 151643
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 23: tokenizer.ggml.add_bos_token bool = false
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 24: tokenizer.chat_template str = {%- if tools %}\n {{- '<|im_start|>...
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 25: general.quantization_version u32 = 2
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 26: split.no u16 = 0
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 27: split.count u16 = 0
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 28: split.tensors.count i32 = 339
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - type f32: 141 tensors
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - type f16: 198 tensors
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_vocab: special tokens cache size = 22
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: time=2024-09-29T16:57:55.318+08:00 level=INFO source=server.go:621 msg="waiting for server to become available" status="llm server loading model"
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_vocab: token to piece cache size = 0.9310 MB
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: format = GGUF V3 (latest)
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: arch = qwen2
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: vocab type = BPE
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_vocab = 152064
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_merges = 151387
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: vocab_only = 0
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_ctx_train = 131072
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_embd = 3584
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_layer = 28
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_head = 28
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_head_kv = 4
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_rot = 128
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_swa = 0
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_embd_head_k = 128
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_embd_head_v = 128
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_gqa = 7
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_embd_k_gqa = 512
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_embd_v_gqa = 512
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: f_norm_eps = 0.0e+00
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: f_norm_rms_eps = 1.0e-06
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: f_clamp_kqv = 0.0e+00
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: f_max_alibi_bias = 0.0e+00
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: f_logit_scale = 0.0e+00
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_ff = 18944
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_expert = 0
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_expert_used = 0
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: causal attn = 1
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: pooling type = 0
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: rope type = 2
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: rope scaling = linear
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: freq_base_train = 1000000.0
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: freq_scale_train = 1
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_ctx_orig_yarn = 131072
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: rope_finetuned = unknown
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: ssm_d_conv = 0
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: ssm_d_inner = 0
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: ssm_d_state = 0
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: ssm_dt_rank = 0
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: ssm_dt_b_c_rms = 0
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: model type = ?B
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: model ftype = F16
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: model params = 7.62 B
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: model size = 14.19 GiB (16.00 BPW)
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: general.name = qwen2.5-7b-instruct
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: BOS token = 151643 '<|endoftext|>'
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: EOS token = 151645 '<|im_end|>'
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: PAD token = 151643 '<|endoftext|>'
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: LF token = 148848 'ÄĬ'
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: EOT token = 151645 '<|im_end|>'
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: max token length = 256
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: ggml_cuda_init: found 1 CUDA devices:
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: Device 0: NVIDIA GeForce RTX 3090, compute capability 8.6, VMM: yes
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_tensors: ggml ctx size = 0.30 MiB
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_tensors: offloading 28 repeating layers to GPU
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_tensors: offloading non-repeating layers to GPU
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_tensors: offloaded 29/29 layers to GPU
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_tensors: CUDA_Host buffer size = 1039.50 MiB
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_tensors: CUDA0 buffer size = 13486.77 MiBDescription
注册编号返回不对
XCF-Mike commented
Model Series
Qwen2.5
What are the models used?
Qwen2.5-7B-Instruct-fp16-gguf
What is the scenario where the problem happened?
ollama
Is this a known issue?
- I have followed the GitHub README.
- I have checked the Qwen documentation and cannot find an answer there.
- I have checked the documentation of the related framework and cannot find useful information.
- I have searched the issues and there is not a similar one.
Information about environment
OS: Ubuntu 22.04 Python: Python 3.11.7 GPUs: 2 * 3090 NVIDIA driver: 535.183.01 CUDA compiler:12.2 PyTorch:none
Log output
9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[2113347]: INFO [main] HTTP server listening | hostname="127.0.0.1" n_threads_http="15" port="44671" tid="129702541623296" timestamp=1727600275 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: loaded meta data with 29 key-value pairs and 339 tensors from /home/ollama/.ollama/models/blobs/sha256-efa98abc3476e6f28591527328e4045e6708c0b0bcf4b3ae836c5cc0435976d8 (version GGUF V3 (latest)) 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 0: general.architecture str = qwen2 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 1: general.type str = model 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 2: general.name str = qwen2.5-7b-instruct 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 3: general.version str = v0.1 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 4: general.finetune str = qwen2.5-7b-instruct 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 5: general.size_label str = 7.6B 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 6: qwen2.block_count u32 = 28 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 7: qwen2.context_length u32 = 131072 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 8: qwen2.embedding_length u32 = 3584 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 9: qwen2.feed_forward_length u32 = 18944 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 10: qwen2.attention.head_count u32 = 28 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 11: qwen2.attention.head_count_kv u32 = 4 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 12: qwen2.rope.freq_base f32 = 1000000.000000 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 13: qwen2.attention.layer_norm_rms_epsilon f32 = 0.000001 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 14: general.file_type u32 = 1 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 15: tokenizer.ggml.model str = gpt2 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 16: tokenizer.ggml.pre str = qwen2 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 17: tokenizer.ggml.tokens arr[str,152064] = ["!", "\"", "#", "$", "%", "&", "'", ... 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 18: tokenizer.ggml.token_type arr[i32,152064] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 19: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 20: tokenizer.ggml.eos_token_id u32 = 151645 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 21: tokenizer.ggml.padding_token_id u32 = 151643 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 22: tokenizer.ggml.bos_token_id u32 = 151643 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 23: tokenizer.ggml.add_bos_token bool = false 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 24: tokenizer.chat_template str = {%- if tools %}\n {{- '<|im_start|>... 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 25: general.quantization_version u32 = 2 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 26: split.no u16 = 0 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 27: split.count u16 = 0 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - kv 28: split.tensors.count i32 = 339 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - type f32: 141 tensors 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llama_model_loader: - type f16: 198 tensors 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_vocab: special tokens cache size = 22 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: time=2024-09-29T16:57:55.318+08:00 level=INFO source=server.go:621 msg="waiting for server to become available" status="llm server loading model" 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_vocab: token to piece cache size = 0.9310 MB 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: format = GGUF V3 (latest) 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: arch = qwen2 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: vocab type = BPE 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_vocab = 152064 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_merges = 151387 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: vocab_only = 0 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_ctx_train = 131072 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_embd = 3584 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_layer = 28 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_head = 28 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_head_kv = 4 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_rot = 128 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_swa = 0 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_embd_head_k = 128 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_embd_head_v = 128 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_gqa = 7 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_embd_k_gqa = 512 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_embd_v_gqa = 512 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: f_norm_eps = 0.0e+00 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: f_norm_rms_eps = 1.0e-06 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: f_clamp_kqv = 0.0e+00 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: f_max_alibi_bias = 0.0e+00 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: f_logit_scale = 0.0e+00 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_ff = 18944 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_expert = 0 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_expert_used = 0 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: causal attn = 1 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: pooling type = 0 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: rope type = 2 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: rope scaling = linear 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: freq_base_train = 1000000.0 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: freq_scale_train = 1 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: n_ctx_orig_yarn = 131072 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: rope_finetuned = unknown 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: ssm_d_conv = 0 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: ssm_d_inner = 0 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: ssm_d_state = 0 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: ssm_dt_rank = 0 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: ssm_dt_b_c_rms = 0 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: model type = ?B 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: model ftype = F16 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: model params = 7.62 B 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: model size = 14.19 GiB (16.00 BPW) 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: general.name = qwen2.5-7b-instruct 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: BOS token = 151643 '<|endoftext|>' 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: EOS token = 151645 '<|im_end|>' 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: PAD token = 151643 '<|endoftext|>' 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: LF token = 148848 'ÄĬ' 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: EOT token = 151645 '<|im_end|>' 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_print_meta: max token length = 256 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: ggml_cuda_init: found 1 CUDA devices: 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: Device 0: NVIDIA GeForce RTX 3090, compute capability 8.6, VMM: yes 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_tensors: ggml ctx size = 0.30 MiB 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_tensors: offloading 28 repeating layers to GPU 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_tensors: offloading non-repeating layers to GPU 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_tensors: offloaded 29/29 layers to GPU 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_tensors: CUDA_Host buffer size = 1039.50 MiB 9月 29 16:57:55 zyzc-Z590-GAMING-X ollama[758604]: llm_load_tensors: CUDA0 buffer size = 13486.77 MiBDescription
注册编号返回不对
请问这个问题现在有解决吗
daimashenjing commented
应该没有解决，我发现 https://huggingface.co/spaces/Qwen/Qwen2.5 上也是会出现一样的结果 只是 注册编号要做一些修改
下面是示例
# 文档：
使用有效期：2022年12月22日
-2025年12月20日
中华人民共和国二级建造师注册证书
名：曹国舅测试
姓别：男
性出生日期：1972-04-08
注册编号：川251200728866335
聘用企业：测试企业测试公司测试测试测试
注册专业：建筑工程(有效期：2022-12-21至2025-12-20)
圆方
省建住房
建设厅
租城
个人签名：
签发日期：2022年12月21日
请登录“四川建设发布”
微信公众号扫一扫查询
签名日期：
# 提取示例的与说明：
{useEndDate=使用有效期,personName=姓名,gender=性别,certNo=注册编号,corpName=聘用企业,array=[regMajor=专业名称,endDate=有效期范围]}
# 提取信息规则：
1.回答的内容请以JSON的格式给出,确保JSON格式准确无误，避免插入任何注释。
2.答案必须严格遵循文档内容即使与事实不符，一定不要自行假设数据或修改，一定不要做多余解释，此为最高原则。
jklj077 commented
cc: @yubowen-ph @yangapku
github-actions commented
This issue has been automatically marked as inactive due to lack of recent activity. Should you believe it remains unresolved and warrants attention, kindly leave a comment on this thread.
daimashenjing commented
还是没有解决
xliber commented
vllm的话，把temperature设置为了0，可以解决这个问题。
jklj077 commented
try this:
你是一名档案整理员，负责从文档中提取结构化信息。你需要遵循以下提取信息的规则：
1. 答案必须严格遵循文档内容。即使与事实不符，一定不要自行假设数据或修改，一定不要做多余解释，此为最高原则。
2. 回答的内容请以JSON的格式给出，确保JSON格式准确无误，避免插入任何注释。
例如，针对以下文档
```
使用有效期：2022年12月22日
-2025年12月20日
中华人民共和国二级建造师注册证书
名：曹国舅测试
姓别：男
性出生日期：1972-04-08
注册编号：川253400728866335
聘用企业：测试企业测试公司测试测试测试
注册专业：建筑工程(有效期：2022-12-21至2025-12-20)
圆方
省建住房
建设厅
租城
个人签名：
签发日期：2022年12月21日
请登录“四川建设发布”
微信公众号扫一扫查询
签名日期：
```
你应返回
```json
{
"useEndDate": "2025年12月20日",
"personName": "曹国舅测试",
"gender": "男",
"certNo": "川253400728866335",
"corpName": "测试企业测试公司测试测试测试",
"array": [
{
"regMajor": "建筑工程",
"endDate": "2022-12-21至2025-12-20"
}
]
}
```
给定以下文档，完成该任务：
```
使用有效期：2021年03月12日
-2020年04月10日
中华人民共和国一级建造师注册证书
名：吴仅有
姓别：女
性出生日期：1910-20-18
注册编号：川251200728866335
聘用企业：测试企业测试公司测试测试测试
注册专业：建筑工程(有效期：2021-03-12至2020-04-10)
```
