Qwen 72B GGUF needs manual rope frequency with llama-cpp-python, llama.cpp gets it right

Running the official Qwen 72B GGUF gives no output with tokens bigger then ~2000 tokens, while running the same prompt through llama.cpp with the same settings directly does give output.

I tried reinstalling and building everything from scratch but the problem stays the same. Running the exact same code with Mixtral works fine.

```
llm = LlamaCpp(
model_kwargs={"split_mode": 2, "tensor_split": [40, 60], "offload_kqv": False},
model_path="/models/Liberated-Qwen1.5-72B-Q3_K_M.gguf",
n_gpu_layers=1000,
n_batch=256,
n_threads=20,
verbose=True,
n_ctx=12000,
max_tokens=30000,
temperature=0.7
)

output = llm(
file_content, # Prompt
max_tokens=3200, # Generate up to 32 tokens, set to None to generate up to the end of the context window
stop=["<|im_end|>"], # Stop generating just before the model would generate a new question
echo=True # Echo the prompt back in the output
) # Generate a completion, can also call create_completion
```

```
ggml_init_cublas: GGML_CUDA_FORCE_MMQ: no
ggml_init_cublas: CUDA_USE_TENSOR_CORES: yes
ggml_init_cublas: found 2 CUDA devices:
Device 0: Tesla P40, compute capability 6.1, VMM: yes
Device 1: Tesla P40, compute capability 6.1, VMM: yes
llama_model_loader: loaded meta data with 21 key-value pairs and 963 tensors from /models/qwen1_5-72b-chat-q3_k_m.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 = Qwen1.5-72B-Chat-AWQ-fp16
llama_model_loader: - kv 2: qwen2.block_count u32 = 80
llama_model_loader: - kv 3: qwen2.context_length u32 = 32768
llama_model_loader: - kv 4: qwen2.embedding_length u32 = 8192
llama_model_loader: - kv 5: qwen2.feed_forward_length u32 = 24576
llama_model_loader: - kv 6: qwen2.attention.head_count u32 = 64
llama_model_loader: - kv 7: qwen2.attention.head_count_kv u32 = 64
llama_model_loader: - kv 8: qwen2.attention.layer_norm_rms_epsilon f32 = 0.000001
llama_model_loader: - kv 9: qwen2.rope.freq_base f32 = 1000000.000000
llama_model_loader: - kv 10: qwen2.use_parallel_residual bool = true
llama_model_loader: - kv 11: tokenizer.ggml.model str = gpt2
llama_model_loader: - kv 12: tokenizer.ggml.tokens arr[str,152064] = ["!", "\"", "#", "$", "%", "&", "'", ...
llama_model_loader: - kv 13: tokenizer.ggml.token_type arr[i32,152064] = [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 %}{{'<|im_...
llama_model_loader: - kv 19: general.quantization_version u32 = 2
llama_model_loader: - kv 20: general.file_type u32 = 12
llama_model_loader: - type f32: 401 tensors
llama_model_loader: - type q3_K: 321 tensors
llama_model_loader: - type q4_K: 155 tensors
llama_model_loader: - type q5_K: 85 tensors
llama_model_loader: - type q6_K: 1 tensors
llm_load_vocab: special tokens definition check successful ( 421/152064 ).
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: n_ctx_train = 32768
llm_load_print_meta: n_embd = 8192
llm_load_print_meta: n_head = 64
llm_load_print_meta: n_head_kv = 64
llm_load_print_meta: n_layer = 80
llm_load_print_meta: n_rot = 128
llm_load_print_meta: n_embd_head_k = 128
llm_load_print_meta: n_embd_head_v = 128
llm_load_print_meta: n_gqa = 1
llm_load_print_meta: n_embd_k_gqa = 8192
llm_load_print_meta: n_embd_v_gqa = 8192
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: n_ff = 24576
llm_load_print_meta: n_expert = 0
llm_load_print_meta: n_expert_used = 0
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 = 70B
llm_load_print_meta: model ftype = Q3_K - Medium
llm_load_print_meta: model params = 72.29 B
llm_load_print_meta: model size = 33.45 GiB (3.98 BPW)
llm_load_print_meta: general.name = Qwen1.5-72B-Chat-AWQ-fp16
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_tensors: ggml ctx size = 1.10 MiB
llm_load_tensors: offloading 80 repeating layers to GPU
llm_load_tensors: offloading non-repeating layers to GPU
llm_load_tensors: offloaded 81/81 layers to GPU
llm_load_tensors: CUDA_Split buffer size = 33734.53 MiB
llm_load_tensors: CPU buffer size = 510.47 MiB
llm_load_tensors: CUDA0 buffer size = 12.53 MiB
...................................................................................................
llama_new_context_with_model: n_ctx = 12000
llama_new_context_with_model: freq_base = 10000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init: CUDA_Host KV buffer size = 30000.00 MiB
llama_new_context_with_model: KV self size = 30000.00 MiB, K (f16): 15000.00 MiB, V (f16): 15000.00 MiB
llama_new_context_with_model: CUDA_Host input buffer size = 20.06 MiB
llama_new_context_with_model: CUDA0 compute buffer size = 2092.50 MiB
llama_new_context_with_model: CUDA_Host compute buffer size = 2678.00 MiB
llama_new_context_with_model: graph splits (measure): 162
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 = 1 | VSX = 0 | MATMUL_INT8 = 0 |
Model metadata: {'general.file_type': '12', 'tokenizer.ggml.bos_token_id': '151643', 'general.architecture': 'qwen2', 'general.name': 'Qwen1.5-72B-Chat-AWQ-fp16', 'qwen2.block_count': '80', 'qwen2.context_length': '32768', 'tokenizer.chat_template': "{% for message in messages %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}", 'qwen2.attention.head_count_kv': '64', 'tokenizer.ggml.padding_token_id': '151643', 'qwen2.embedding_length': '8192', 'qwen2.attention.layer_norm_rms_epsilon': '0.000001', 'qwen2.attention.head_count': '64', 'tokenizer.ggml.eos_token_id': '151645', 'qwen2.rope.freq_base': '1000000.000000', 'qwen2.use_parallel_residual': 'true', 'general.quantization_version': '2', 'qwen2.feed_forward_length': '24576', 'tokenizer.ggml.model': 'gpt2'}
```

==============================>

```
llama_print_timings: load time = 12511.09 ms
llama_print_timings: sample time = 3.70 ms / 1 runs ( 3.70 ms per token, 270.05 tokens per second)
llama_print_timings: prompt eval time = 146351.02 ms / 2617 tokens ( 55.92 ms per token, 17.88 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 = 146604.53 ms / 2618 tokens
```