Server UI bug: corrupted generation

[ivanstepanovftw]

What happened?

Server somehow corrupted the prompt, so tokens at the end of the every line are lost.

Here is how I run server:

./build/bin/llama-server -m ~/Downloads/qwen2.5-7b-instruct-q4_0-00001-of-00002.gguf

Here is how I test CLI to ensure it is a server bug:

./build/bin/llama-cli -m ~/Downloads/qwen2.5-7b-instruct-q4_0-00001-of-00002.gguf -e -p "<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n<|im_start|>user\nHi\!<|im_end|>\n<|im_start|>assistant\nHow can I assist you today?<|im_end|>\n<|im_start|>user\nImplement fibbonaci in Python<|im_end|>\n<|im_start|>assistant\n" -n 128 -t 7 -tb 8 --temp 0

Here is the output from the CLI

➜  llama.cpp git:(master) ✗ ./build/bin/llama-cli -m ~/Downloads/qwen2.5-7b-instruct-q4_0-00001-of-00002.gguf -e -p "<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n<|im_start|>user\nHi\!<|im_end|>\n<|im_start|>assistant\nHow can I assist you today?<|im_end|>\n<|im_start|>user\nImplement fibbonaci in Python<|im_end|>\n<|im_start|>assistant\n" -n 128 -t 7 -tb 8 --temp 0          
build: 3891 (d5cb8684) with cc (Debian 12.2.0-14) 12.2.0 for x86_64-linux-gnu
main: llama backend init
main: load the model and apply lora adapter, if any
llama_model_loader: additional 1 GGUFs metadata loaded.
llama_model_loader: loaded meta data with 29 key-value pairs and 339 tensors from /home/i/Downloads/qwen2.5-7b-instruct-q4_0-00001-of-00002.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.type str              = model
llama_model_loader: - kv   2:                               general.name str              = qwen2.5-7b-instruct
llama_model_loader: - kv   3:                            general.version str              = v0.1
llama_model_loader: - kv   4:                           general.finetune str              = qwen2.5-7b-instruct
llama_model_loader: - kv   5:                         general.size_label str              = 7.6B
llama_model_loader: - kv   6:                          qwen2.block_count u32              = 28
llama_model_loader: - kv   7:                       qwen2.context_length u32              = 131072
llama_model_loader: - kv   8:                     qwen2.embedding_length u32              = 3584
llama_model_loader: - kv   9:                  qwen2.feed_forward_length u32              = 18944
llama_model_loader: - kv  10:                 qwen2.attention.head_count u32              = 28
llama_model_loader: - kv  11:              qwen2.attention.head_count_kv u32              = 4
llama_model_loader: - kv  12:                       qwen2.rope.freq_base f32              = 1000000.000000
llama_model_loader: - kv  13:     qwen2.attention.layer_norm_rms_epsilon f32              = 0.000001
llama_model_loader: - kv  14:                          general.file_type u32              = 2
llama_model_loader: - kv  15:                       tokenizer.ggml.model str              = gpt2
llama_model_loader: - kv  16:                         tokenizer.ggml.pre str              = qwen2
llama_model_loader: - kv  17:                      tokenizer.ggml.tokens arr[str,152064]  = ["!", "\"", "#", "$", "%", "&", "'", ...
llama_model_loader: - kv  18:                  tokenizer.ggml.token_type arr[i32,152064]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv  19:                      tokenizer.ggml.merges arr[str,151387]  = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
llama_model_loader: - kv  20:                tokenizer.ggml.eos_token_id u32              = 151645
llama_model_loader: - kv  21:            tokenizer.ggml.padding_token_id u32              = 151643
llama_model_loader: - kv  22:                tokenizer.ggml.bos_token_id u32              = 151643
llama_model_loader: - kv  23:               tokenizer.ggml.add_bos_token bool             = false
llama_model_loader: - kv  24:                    tokenizer.chat_template str              = {%- if tools %}\n    {{- '<|im_start|>...
llama_model_loader: - kv  25:               general.quantization_version u32              = 2
llama_model_loader: - kv  26:                                   split.no u16              = 0
llama_model_loader: - kv  27:                                split.count u16              = 2
llama_model_loader: - kv  28:                        split.tensors.count i32              = 339
llama_model_loader: - type  f32:  141 tensors
llama_model_loader: - type q4_0:  197 tensors
llama_model_loader: - type q6_K:    1 tensors
llm_load_vocab: special tokens cache size = 22
llm_load_vocab: token to piece cache size = 0.9310 MB
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          = 152064
llm_load_print_meta: n_merges         = 151387
llm_load_print_meta: vocab_only       = 0
llm_load_print_meta: n_ctx_train      = 131072
llm_load_print_meta: n_embd           = 3584
llm_load_print_meta: n_layer          = 28
llm_load_print_meta: n_head           = 28
llm_load_print_meta: n_head_kv        = 4
llm_load_print_meta: n_rot            = 128
llm_load_print_meta: n_swa            = 0
llm_load_print_meta: n_embd_head_k    = 128
llm_load_print_meta: n_embd_head_v    = 128
llm_load_print_meta: n_gqa            = 7
llm_load_print_meta: n_embd_k_gqa     = 512
llm_load_print_meta: n_embd_v_gqa     = 512
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             = 18944
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_ctx_orig_yarn  = 131072
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: ssm_dt_b_c_rms   = 0
llm_load_print_meta: model type       = ?B
llm_load_print_meta: model ftype      = Q4_0
llm_load_print_meta: model params     = 7.62 B
llm_load_print_meta: model size       = 4.12 GiB (4.65 BPW) 
llm_load_print_meta: general.name     = qwen2.5-7b-instruct
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|>'
llm_load_print_meta: EOG token        = 151643 '<|endoftext|>'
llm_load_print_meta: EOG token        = 151645 '<|im_end|>'
llm_load_print_meta: max token length = 256
llm_load_tensors: ggml ctx size =    0.15 MiB
llm_load_tensors:        CPU buffer size =  3793.03 MiB
llm_load_tensors:        CPU buffer size =   427.40 MiB
.....................................................................................
llama_new_context_with_model: n_ctx      = 131072
llama_new_context_with_model: n_batch    = 2048
llama_new_context_with_model: n_ubatch   = 512
llama_new_context_with_model: flash_attn = 0
llama_new_context_with_model: freq_base  = 1000000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init:        CPU KV buffer size =  7168.00 MiB
llama_new_context_with_model: KV self size  = 7168.00 MiB, K (f16): 3584.00 MiB, V (f16): 3584.00 MiB
llama_new_context_with_model:        CPU  output buffer size =     0.58 MiB
llama_new_context_with_model:        CPU compute buffer size =  7452.01 MiB
llama_new_context_with_model: graph nodes  = 986
llama_new_context_with_model: graph splits = 1
llama_init_from_gpt_params: warming up the model with an empty run - please wait ... (--no-warmup to disable)
main: llama threadpool init, n_threads = 7

system_info: n_threads = 7 (n_threads_batch = 8) / 16 | 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 | RISCV_VECT = 0 | WASM_SIMD = 0 | BLAS = 0 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 | 

sampler seed: 4294967295
sampler params: 
	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 = 0.000
	mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000
sampler chain: logits -> logit-bias -> penalties -> greedy 
generate: n_ctx = 131072, n_batch = 2048, n_predict = 128, n_keep = 0

system
You are a helpful assistant.
user
Hi!
assistant
How can I assist you today?
user
Implement fibbonaci in Python
assistant
Sure! Here are a few ways to implement the Fibonacci sequence in Python:

1. **Iterative Approach:**
   ```python
   def fibonacci(n):
       if n <= 0:
           return []
       elif n == 1:
           return [0]
       elif n == 2:
           return [0, 1]
       
       fib_sequence = [0, 1]
       for i in range(2, n):
           next_value = fib_sequence[-1] + fib_sequence[-2]
           fib_sequence.append(next_value)
       return fib_sequence

   # Example usage
   print(fibonacci(

llama_perf_sampler_print:    sampling time =      25.14 ms /   172 runs   (    0.15 ms per token,  6840.33 tokens per second)
llama_perf_context_print:        load time =   27227.67 ms
llama_perf_context_print: prompt eval time =    6480.76 ms /    44 tokens (  147.29 ms per token,     6.79 tokens per second)
llama_perf_context_print:        eval time =   20080.14 ms /   127 runs   (  158.11 ms per token,     6.32 tokens per second)
llama_perf_context_print:       total time =   26704.27 ms /   171 tokens
Time: 0h:00m:56s                                                                                                                                                
➜  llama.cpp git:(master) ✗ 

Here is how I test server endpoints to ensure this is a UI bug:

import httpx

# Define the URL and the headers
url = 'http://localhost:8080/completion'
headers = {
    'Content-Type': 'application/json'
}

# Define the JSON payload with properly escaped newlines
data = {
    "prompt": "<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n<|im_start|>user\nHi!<|im_end|>\n<|im_start|>assistant\nHow can I assist you today?<|im_end|>\n<|im_start|>user\nImplement fibbonaci in Python<|im_end|>\n<|im_start|>assistant\n",
    "n_predict": 128
}

# Send the POST request using httpx with no timeout
response = httpx.post(url, json=data, headers=headers, timeout=None)

# Print the response from the server
print(response.json())

Response from the endpoints are valid:

{'content': 'Sure! Here are a few ways to implement the Fibonacci sequence in Python:\n\n1. **Iterative Approach:**\n   ```python\n   def fibonacci(n):\n       a, b = 0, 1\n       for _ in range(n):\n           a, b = b, a + b\n       return a\n\n   # Example usage\n   n = 10\n   print(f"Fibonacci({n}) = {fibonacci(n)}")\n   ```\n\n2. **Recursive Approach:**\n   ```python\n   def fibonacci(n):\n       if n <= 0:\n           return 0\n       elif n ==', 'id_slot': 0, 'stop': True, 'model': '/home/i/Downloads/qwen2.5-7b-instruct-q4_0-00001-of-00002.gguf', 'tokens_predicted': 128, 'tokens_evaluated': 44, 'generation_settings': {'n_ctx': 131072, 'n_predict': -1, 'model': '/home/i/Downloads/qwen2.5-7b-instruct-q4_0-00001-of-00002.gguf', 'seed': 4294967295, 'seed_cur': 3124811782, 'temperature': 0.800000011920929, 'dynatemp_range': 0.0, 'dynatemp_exponent': 1.0, 'top_k': 40, 'top_p': 0.949999988079071, 'min_p': 0.05000000074505806, 'tfs_z': 1.0, 'typical_p': 1.0, 'repeat_last_n': 64, 'repeat_penalty': 1.0, 'presence_penalty': 0.0, 'frequency_penalty': 0.0, 'mirostat': 0, 'mirostat_tau': 5.0, 'mirostat_eta': 0.10000000149011612, 'penalize_nl': False, 'stop': [], 'max_tokens': 128, 'n_keep': 0, 'n_discard': 0, 'ignore_eos': False, 'stream': False, 'n_probs': 0, 'min_keep': 0, 'grammar': '', 'samplers': ['top_k', 'tfs_z', 'typ_p', 'top_p', 'min_p', 'temperature']}, 'prompt': '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n<|im_start|>user\nHi!<|im_end|>\n<|im_start|>assistant\nHow can I assist you today?<|im_end|>\n<|im_start|>user\nImplement fibbonaci in Python<|im_end|>\n<|im_start|>assistant\n', 'truncated': False, 'stopped_eos': False, 'stopped_word': False, 'stopped_limit': True, 'stopping_word': '', 'tokens_cached': 171, 'timings': {'prompt_n': 44, 'prompt_ms': 2533.391, 'prompt_per_token_ms': 57.577068181818184, 'prompt_per_second': 17.368025701520214, 'predicted_n': 128, 'predicted_ms': 17878.5, 'predicted_per_token_ms': 139.67578125, 'predicted_per_second': 7.159437312973684}, 'index': 0}

Here are screenshots:

Old web UI

New web UI

New web UI Chat

SimpleChat

llama-cli

What is affected:

• server ui
• server new ui

Unaffected:

• server endpoints
• server SimpleChat
• CLI

Name and Version

version: 3891 (d5cb868)
built with cc (Debian 12.2.0-14) 12.2.0 for x86_64-linux-gnu

What operating system are you seeing the problem on?

No response

Relevant log output

No response