| Padding must be 'center' or 'same'. | validation | error | |
| `penalty` has to be a strictly positive float, but is {penal | validation | error | |
| Out of memory! No free blocks are available. | validation | error | |
| Double free! {block} is already freed. | validation | error | |
| Unknown load format: {self.load_format}. Must be one of 'aut | validation | error | |
| load format '{load_format}' is not supported in ROCm. Suppor | validation | error | |
| Currently, the 'pt' format is not supported for Mixtral. Ple | validation | error | |
| Unknown tokenizer mode: {self.tokenizer_mode}. Must be eithe | validation | error | |
| Quantization method specified in the model config ({hf_quant | validation | error | |
| Unknown quantization method: {self.quantization}. Must be on | validation | error | |
| {self.quantization} quantization is currently not supported | validation | error | |
| Total number of attention heads ({total_num_attention_heads} | validation | error | |
| Total number of hidden layers ({total_num_hidden_layers}) mu | validation | error | |
| GPU memory utilization must be less than 1.0. Got {self.gpu_ | validation | error | |
| Too large swap space. {msg} | validation | error | |
| Pipeline parallelism is not supported yet. | exception | error | |
| max_num_batched_tokens ({self.max_num_batched_tokens}) is sm | validation | error | |
| max_num_batched_tokens ({self.max_num_batched_tokens}) must | validation | error | |
| Unknown dtype: {dtype} | validation | error | |
| dtype '{dtype}' is not supported in ROCm. Supported dtypes a | validation | error | rocm, dtype, amd, gpu, configuration |
| User-specified max_model_len ({max_model_len}) is greater th | validation | error | max-model-len, context-length, rope-scaling, configuration |
| Unsupported activation: {hidden_act}. Only silu is supported | validation | error | activation, silu, llama, model-config |
| Either prompts or prompt_token_ids must be provided. | validation | error | api-misuse, input-validation, generate |
| The lengths of prompts and prompt_token_ids must be the same | validation | error | api-misuse, input-validation, length-mismatch |
| Ray does not allocate any GPUs on the driver node. Consider | validation | critical | ray, distributed, gpu-scheduling, placement-group |
| The model's max seq len ({self.model_config.max_model_len}) | validation | critical | gpu-memory, kv-cache, max-model-len, context-length |
| max_concurrent_workers is not supported yet. | exception | error | not-implemented, vllm, internal-api |
| The quantization method {model_config.quantization} is not s | validation | error | quantization, awq, gptq, cuda-capability, gpu |
| {model_config.dtype} is not supported for quantization metho | validation | error | quantization, dtype, vllm, gpu, model-loading |
| n must be at least 1, got {self.n}. | validation | error | sampling-params, validation, llm-generation, constructor |
| best_of must be greater than or equal to n, got n={self.n} a | validation | error | sampling-params, validation, llm-generation, constructor |
| presence_penalty must be in [-2, 2], got {self.presence_pena | validation | error | sampling-params, validation, presence-penalty, llm-generation |
| frequency_penalty must be in [-2, 2], got {self.frequency_pe | validation | error | sampling-params, validation, frequency-penalty, llm-generation |
| repetition_penalty must be in (0, 2], got {self.repetition_p | validation | error | sampling-params, validation, repetition-penalty, llm-generation |
| top_p must be in (0, 1], got {self.top_p}. | validation | error | sampling-params, validation, top-p, nucleus-sampling |
| top_k must be -1 (disable), or at least 1, got {self.top_k}. | validation | error | sampling-params, validation, top-k, llm-generation |
| min_p must be in [0, 1], got {self.min_p}. | validation | error | sampling-params, validation, min-p, llm-generation |
| max_tokens must be at least 1, got {self.max_tokens}. | validation | error | sampling-params, validation, max-tokens, llm-generation |