qwen 1.5 Beta 1.8B output incoherently

[sorasoras]

latest llama cpp output incoherently compare to Transformers output.

transformers/vllm work ok but llama cpp gguf does not

[zhengxingmao]

+1 Both Qwen1.5-72B-Chat and Qwen-72B-Chat output incoherently. The old llama_cpp which nearly 2023 Dec worked normal.

[sorasoras]

+1 Both Qwen1.5-72B-Chat and Qwen-72B-Chat output incoherently. The old llama_cpp which nearly 2023 Dec worked normal.

That's Great info to know. Can you pinpoint which version is the last version work ? If we can pinpoint which change is the cause of incoherence, it might get us close to solving the problem.

[riverzhou]

Same problem.

[zhengxingmao]

There are some mistakes in model config files . I used the Qwen1.5'gguf from huaggingface which run successfully. Maybe relate to this PR https://huggingface.co/Qwen/Qwen1.5-72B-Chat/commit/bc11a298a0c6a5cd737064db62c6ad20ec6331be

[RonanKMcGovern]

Hmm, I'm unsure that's the only issue. I chat-fine tuned and tried to quantize since then.

…

On Fri, Mar 1, 2024 at 1:36 AM weimy ***@***.***> wrote: There are some mistake in model config files . I used the Qwen1.5'gguf from huaggingface which run successfully. Maybe relate to this PR https://huggingface.co/Qwen/Qwen1.5-72B-Chat/commit/bc11a298a0c6a5cd737064db62c6ad20ec6331be — Reply to this email directly, view it on GitHub <#5459 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ASVG6CVVN3PLRQJLRZ75QTDYV7LRFAVCNFSM6AAAAABDEP5PX6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNZSGI4DQMBWGM> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[sorasoras]

Hmm, I'm unsure that's the only issue. I chat-fine tuned and tried to quantize since then.
…
On Fri, Mar 1, 2024 at 1:36 AM weimy @.> wrote: There are some mistake in model config files . I used the Qwen1.5'gguf from huaggingface which run successfully. Maybe relate to this PR https://huggingface.co/Qwen/Qwen1.5-72B-Chat/commit/bc11a298a0c6a5cd737064db62c6ad20ec6331be — Reply to this email directly, view it on GitHub <#5459 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ASVG6CVVN3PLRQJLRZ75QTDYV7LRFAVCNFSM6AAAAABDEP5PX6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNZSGI4DQMBWGM . You are receiving this because you are subscribed to this thread.Message ID: @.>

mostly,but there might me some conf need to adjust in EOS in the conf of the original model.

[NineMeowICT]

So, is this problem solved?

[sorasoras]

So, is this problem solved?

Not in the official repo

[ChaoII]

I have the same problem.

(llama) D:\llama.cpp\build\install\bin>main.exe -m D:/Qwen1.5-0.5B-Chat/ggml-model-f16.gguf -p "What's your name?"
Log start
main: build = 2725 (784e11de)
main: built with MSVC 19.35.32215.0 for
main: seed  = 1714032293
llama_model_loader: loaded meta data with 19 key-value pairs and 291 tensors from D:/Qwen1.5-0.5B-Chat/ggml-model-f16.gguf (version GGUF V3 (latest))
llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
llama_model_loader: - kv   0:                       general.architecture str              = qwen2
llama_model_loader: - kv   1:                               general.name str              = Qwen1.5-0.5B-Chat
llama_model_loader: - kv   2:                          qwen2.block_count u32              = 24
llama_model_loader: - kv   3:                       qwen2.context_length u32              = 32768
llama_model_loader: - kv   4:                     qwen2.embedding_length u32              = 1024
llama_model_loader: - kv   5:                  qwen2.feed_forward_length u32              = 2816
llama_model_loader: - kv   6:                 qwen2.attention.head_count u32              = 16
llama_model_loader: - kv   7:              qwen2.attention.head_count_kv u32              = 16
llama_model_loader: - kv   8:                       qwen2.rope.freq_base f32              = 1000000.000000
llama_model_loader: - kv   9:     qwen2.attention.layer_norm_rms_epsilon f32              = 0.000001
llama_model_loader: - kv  10:                          general.file_type u32              = 1
llama_model_loader: - kv  11:                       tokenizer.ggml.model str              = gpt2
llama_model_loader: - kv  12:                      tokenizer.ggml.tokens arr[str,151936]  = ["!", "\"", "#", "$", "%", "&", "'", ...
llama_model_loader: - kv  13:                  tokenizer.ggml.token_type arr[i32,151936]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv  14:                      tokenizer.ggml.merges arr[str,151387]  = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
llama_model_loader: - kv  15:                tokenizer.ggml.eos_token_id u32              = 151645
llama_model_loader: - kv  16:            tokenizer.ggml.padding_token_id u32              = 151643
llama_model_loader: - kv  17:                tokenizer.ggml.bos_token_id u32              = 151643
llama_model_loader: - kv  18:                    tokenizer.chat_template str              = {% for message in messages %}{% if lo...
llama_model_loader: - type  f32:  121 tensors
llama_model_loader: - type  f16:  170 tensors
llm_load_vocab: special tokens definition check successful ( 293/151936 ).
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = qwen2
llm_load_print_meta: vocab type       = BPE
llm_load_print_meta: n_vocab          = 151936
llm_load_print_meta: n_merges         = 151387
llm_load_print_meta: n_ctx_train      = 32768
llm_load_print_meta: n_embd           = 1024
llm_load_print_meta: n_head           = 16
llm_load_print_meta: n_head_kv        = 16
llm_load_print_meta: n_layer          = 24
llm_load_print_meta: n_rot            = 64
llm_load_print_meta: n_embd_head_k    = 64
llm_load_print_meta: n_embd_head_v    = 64
llm_load_print_meta: n_gqa            = 1
llm_load_print_meta: n_embd_k_gqa     = 1024
llm_load_print_meta: n_embd_v_gqa     = 1024
llm_load_print_meta: f_norm_eps       = 0.0e+00
llm_load_print_meta: f_norm_rms_eps   = 1.0e-06
llm_load_print_meta: f_clamp_kqv      = 0.0e+00
llm_load_print_meta: f_max_alibi_bias = 0.0e+00
llm_load_print_meta: f_logit_scale    = 0.0e+00
llm_load_print_meta: n_ff             = 2816
llm_load_print_meta: n_expert         = 0
llm_load_print_meta: n_expert_used    = 0
llm_load_print_meta: causal attn      = 1
llm_load_print_meta: pooling type     = 0
llm_load_print_meta: rope type        = 2
llm_load_print_meta: rope scaling     = linear
llm_load_print_meta: freq_base_train  = 1000000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_yarn_orig_ctx  = 32768
llm_load_print_meta: rope_finetuned   = unknown
llm_load_print_meta: ssm_d_conv       = 0
llm_load_print_meta: ssm_d_inner      = 0
llm_load_print_meta: ssm_d_state      = 0
llm_load_print_meta: ssm_dt_rank      = 0
llm_load_print_meta: model type       = 0.5B
llm_load_print_meta: model ftype      = F16
llm_load_print_meta: model params     = 619.57 M
llm_load_print_meta: model size       = 1.15 GiB (16.00 BPW)
llm_load_print_meta: general.name     = Qwen1.5-0.5B-Chat
llm_load_print_meta: BOS token        = 151643 '<|endoftext|>'
llm_load_print_meta: EOS token        = 151645 '<|im_end|>'
llm_load_print_meta: PAD token        = 151643 '<|endoftext|>'
llm_load_print_meta: LF token         = 148848 'ÄĬ'
llm_load_print_meta: EOT token        = 151645 '<|im_end|>'
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:   no
ggml_cuda_init: CUDA_USE_TENSOR_CORES: yes
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 4060 Ti, compute capability 8.9, VMM: yes
llm_load_tensors: ggml ctx size =    0.14 MiB
llm_load_tensors: offloading 0 repeating layers to GPU
llm_load_tensors: offloaded 0/25 layers to GPU
llm_load_tensors:        CPU buffer size =  1181.97 MiB
....................................................
llama_new_context_with_model: n_ctx      = 512
llama_new_context_with_model: n_batch    = 512
llama_new_context_with_model: n_ubatch   = 512
llama_new_context_with_model: freq_base  = 1000000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init:  CUDA_Host KV buffer size =    48.00 MiB
llama_new_context_with_model: KV self size  =   48.00 MiB, K (f16):   24.00 MiB, V (f16):   24.00 MiB
llama_new_context_with_model:  CUDA_Host  output buffer size =     0.58 MiB
llama_new_context_with_model:      CUDA0 compute buffer size =   595.50 MiB
llama_new_context_with_model:  CUDA_Host compute buffer size =     5.01 MiB
llama_new_context_with_model: graph nodes  = 846
llama_new_context_with_model: graph splits = 340

system_info: n_threads = 10 / 20 | AVX = 1 | AVX_VNNI = 0 | AVX2 = 1 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | FMA = 1 | NEON = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 0 | VSX = 0 | MATMUL_INT8 = 0 |
sampling:
        repeat_last_n = 64, repeat_penalty = 1.000, frequency_penalty = 0.000, presence_penalty = 0.000
        top_k = 40, tfs_z = 1.000, top_p = 0.950, min_p = 0.050, typical_p = 1.000, temp = 0.800
        mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000
sampling order:
CFG -> Penalties -> top_k -> tfs_z -> typical_p -> top_p -> min_p -> temperature
generate: n_ctx = 512, n_batch = 2048, n_predict = -1, n_keep = 0


What's your name?<|im_end|> [end of text]

llama_print_timings:        load time =     361.57 ms
llama_print_timings:      sample time =       0.08 ms /     1 runs   (    0.08 ms per token, 12195.12 tokens per second)
llama_print_timings: prompt eval time =      37.07 ms /     5 tokens (    7.41 ms per token,   134.88 tokens per second)
llama_print_timings:        eval time =       0.00 ms /     1 runs   (    0.00 ms per token,      inf tokens per second)
llama_print_timings:       total time =      38.76 ms /     6 tokens
Log end

[github-actions]

This issue was closed because it has been inactive for 14 days since being marked as stale.