search

RasaHQ / rasa-calm-demo

52 stars 18 forks source link

Integration of Llama with Rasa Pro #35

Closed hamzaziizzz closed 2 months ago

hamzaziizzz commented 5 months ago

I am geeting the following for Llama-LLM

2024-06-28 21:57:20 INFO     openai  - message='OpenAI API response' path=https://api.openai.com/v1/embeddings processing_ms=15 request_id=req_5533598d1e5469fd213c359055ce074d response_code=200
llama_model_loader: loaded meta data with 23 key-value pairs and 543 tensors from models/llava-v1.6-34b.Q6_K.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              = llama
llama_model_loader: - kv   1:                               general.name str              = LLaMA v2
llama_model_loader: - kv   2:                       llama.context_length u32              = 4096
llama_model_loader: - kv   3:                     llama.embedding_length u32              = 7168
llama_model_loader: - kv   4:                          llama.block_count u32              = 60
llama_model_loader: - kv   5:                  llama.feed_forward_length u32              = 20480
llama_model_loader: - kv   6:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv   7:                 llama.attention.head_count u32              = 56
llama_model_loader: - kv   8:              llama.attention.head_count_kv u32              = 8
llama_model_loader: - kv   9:     llama.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  10:                       llama.rope.freq_base f32              = 5000000.000000
llama_model_loader: - kv  11:                          general.file_type u32              = 18
llama_model_loader: - kv  12:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  13:                      tokenizer.ggml.tokens arr[str,64000]   = ["<unk>", "<|startoftext|>", "<|endof...
llama_model_loader: - kv  14:                      tokenizer.ggml.scores arr[f32,64000]   = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  15:                  tokenizer.ggml.token_type arr[i32,64000]   = [2, 3, 3, 3, 3, 3, 1, 1, 1, 3, 3, 3, ...
llama_model_loader: - kv  16:                tokenizer.ggml.bos_token_id u32              = 1
llama_model_loader: - kv  17:                tokenizer.ggml.eos_token_id u32              = 7
llama_model_loader: - kv  18:            tokenizer.ggml.padding_token_id u32              = 0
llama_model_loader: - kv  19:               tokenizer.ggml.add_bos_token bool             = false
llama_model_loader: - kv  20:               tokenizer.ggml.add_eos_token bool             = false
llama_model_loader: - kv  21:                    tokenizer.chat_template str              = {% for message in messages %}{{'<|im_...
llama_model_loader: - kv  22:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:  121 tensors
llama_model_loader: - type q6_K:  422 tensors
llm_load_vocab: special tokens cache size = 267
llm_load_vocab: token to piece cache size = 0.3834 MB
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = llama
llm_load_print_meta: vocab type       = SPM
llm_load_print_meta: n_vocab          = 64000
llm_load_print_meta: n_merges         = 0
llm_load_print_meta: n_ctx_train      = 4096
llm_load_print_meta: n_embd           = 7168
llm_load_print_meta: n_head           = 56
llm_load_print_meta: n_head_kv        = 8
llm_load_print_meta: n_layer          = 60
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            = 7
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-05
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             = 20480
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        = 0
llm_load_print_meta: rope scaling     = linear
llm_load_print_meta: freq_base_train  = 5000000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_ctx_orig_yarn  = 4096
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       = 30B
llm_load_print_meta: model ftype      = Q6_K
llm_load_print_meta: model params     = 34.39 B
llm_load_print_meta: model size       = 26.27 GiB (6.56 BPW) 
llm_load_print_meta: general.name     = LLaMA v2
llm_load_print_meta: BOS token        = 1 '<|startoftext|>'
llm_load_print_meta: EOS token        = 7 '<|im_end|>'
llm_load_print_meta: UNK token        = 0 '<unk>'
llm_load_print_meta: PAD token        = 0 '<unk>'
llm_load_print_meta: LF token         = 315 '<0x0A>'
llm_load_print_meta: EOT token        = 2 '<|endoftext|>'
llm_load_tensors: ggml ctx size =    0.28 MiB
llm_load_tensors:        CPU buffer size = 26905.46 MiB
....................................................................................................
llama_new_context_with_model: n_batch is less than GGML_KQ_MASK_PAD - increasing to 32
llama_new_context_with_model: n_ctx      = 512
llama_new_context_with_model: n_batch    = 32
llama_new_context_with_model: n_ubatch   = 32
llama_new_context_with_model: flash_attn = 0
llama_new_context_with_model: freq_base  = 10000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init:        CPU KV buffer size =   120.00 MiB
llama_new_context_with_model: KV self size  =  120.00 MiB, K (f16):   60.00 MiB, V (f16):   60.00 MiB
llama_new_context_with_model:        CPU  output buffer size =     0.24 MiB
llama_new_context_with_model:        CPU compute buffer size =     8.69 MiB
llama_new_context_with_model: graph nodes  = 1926
llama_new_context_with_model: graph splits = 1
AVX = 1 | AVX_VNNI = 1 | 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 | WASM_SIMD = 0 | BLAS = 0 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 | 
Model metadata: {'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 %}", 'tokenizer.ggml.add_eos_token': 'false', 'tokenizer.ggml.padding_token_id': '0', 'tokenizer.ggml.eos_token_id': '7', 'general.architecture': 'llama', 'llama.rope.freq_base': '5000000.000000', 'llama.context_length': '4096', 'general.name': 'LLaMA v2', 'tokenizer.ggml.add_bos_token': 'false', 'llama.embedding_length': '7168', 'llama.feed_forward_length': '20480', 'llama.attention.layer_norm_rms_epsilon': '0.000010', 'llama.rope.dimension_count': '128', 'tokenizer.ggml.bos_token_id': '1', 'llama.attention.head_count': '56', 'llama.block_count': '60', 'llama.attention.head_count_kv': '8', 'general.quantization_version': '2', 'tokenizer.ggml.model': 'llama', 'general.file_type': '18'}
Available chat formats from metadata: chat_template.default
Guessed chat format: chatml
2024-06-28 21:57:56 ERROR    rasa.dialogue_understanding.generator.llm_command_generator  - [error    ] llm_command_generator.llm.error error=ValueError('Requested tokens (804) exceed context window of 512')
/home/hamza/PycharmProjects/G6-Voice-Assistant/.venv/lib/python3.10/site-packages/sanic/server/websockets/impl.py:521: DeprecationWarning: The explicit passing of coroutine objects to asyncio.wait() is deprecated since Python 3.8, and scheduled for removal in Python 3.11.

My config.yml is as follows

recipe: default.v1
language: en
pipeline:
- name: LLMCommandGenerator
  llm:
    type: llamacpp
    model_path: "models/llava-v1.6-34b.Q6_K.gguf"
    chunk_size: 16
    model_kwargs:
      device: "gpu"
#  llm:
#    model_name: gpt-4

policies:
- name: FlowPolicy
#  - name: EnterpriseSearchPolicy
#  - name: RulePolicy
assistant_id: 20240627-152245-khaki-isotope
SanjuktaK commented 4 months ago

are you able to integrate it?

maksim-m commented 2 months ago

Hey @hamzaziizzz and @SanjuktaK,

In the Rasa Pro 3.10 release, we've improved the configuration process for self-hosted LLMs. To use a self-hosted LLM, you need to set provider: self-hosted and specify the correct URL for your OpenAI-compatible API endpoint using the api_base key. Here’s an example configuration:

- name: SingleStepLLMCommandGenerator
  llm:
    provider: self-hosted
    model: meta-llama/CodeLlama-7b-Instruct-hf
    api_base: "https://my-endpoint/v1"

Make sure you're using Rasa Pro version 3.10 or higher. You can find more details in our documentation: Rasa Pro - LLM Configuration - Self Hosted Model Server.

As for the Llama model itself, I recently tested one in the GGUF format, and it worked seamlessly with CALM. Here’s the configuration I used:

- name: SingleStepLLMCommandGenerator
  llm:
    provider: "self-hosted"
    model: "ggml-org/Meta-Llama-3.1-8B-Instruct-Q4_0-GGUF"
    api_base: "http://localhost:8080/v1"
    request_timeout: 30
    temperature: 0.0
    top_p: 0.0

I hope this helps.