Grok 1.0 Model Stats

xAI’s Grok 1.0 model (see Github repo) has 64 layers, 8K context length, in total 314B parameters.

The model architecture is Mixture of 8 Experts (MoE), with 48 attention heads for queries, 8 attention heads for keys/values, and 6,144 embedding dimension.

Notably, the model uses a vocabulary of 131,072 tokens in its tokenizer. Comparatively, mixtral and llama-2 models use only 32000 tokens.

Here is the more info:

llama_model_loader: - kv   0:                       general.architecture str              = grok
llama_model_loader: - kv   1:                               general.name str              = Grok
llama_model_loader: - kv   2:                           grok.block_count u32              = 64
llama_model_loader: - kv   3:                        grok.context_length u32              = 8192
llama_model_loader: - kv   4:                      grok.embedding_length u32              = 6144
llama_model_loader: - kv   5:                   grok.feed_forward_length u32              = 32768
llama_model_loader: - kv   6:                  grok.attention.head_count u32              = 48
llama_model_loader: - kv   7:               grok.attention.head_count_kv u32              = 8
llama_model_loader: - kv   8:                        grok.rope.freq_base f32              = 10000.000000
llama_model_loader: - kv   9:      grok.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  10:                          grok.expert_count u32              = 8
llama_model_loader: - kv  11:                     grok.expert_used_count u32              = 2
llama_model_loader: - kv  12:                          general.file_type u32              = 22
llama_model_loader: - kv  13:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  14:                      tokenizer.ggml.tokens arr[str,131072]  = ["[PAD]", "[BOS]", "[EOS]", "[UNK]", ...
llama_model_loader: - kv  15:                      tokenizer.ggml.scores arr[f32,131072]  = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  16:                  tokenizer.ggml.token_type arr[i32,131072]  = [3, 3, 3, 2, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv  17:                tokenizer.ggml.bos_token_id u32              = 1
llama_model_loader: - kv  18:                tokenizer.ggml.eos_token_id u32              = 2
llama_model_loader: - kv  19:            tokenizer.ggml.padding_token_id u32              = 0
llama_model_loader: - kv  20:               tokenizer.ggml.add_bos_token bool             = false
llama_model_loader: - kv  21:               tokenizer.ggml.add_eos_token bool             = false
llama_model_loader: - kv  22:               general.quantization_version u32              = 2
llama_model_loader: - kv  23:                                   split.no u16              = 0
llama_model_loader: - kv  24:                                split.count u16              = 9
llama_model_loader: - kv  25:                        split.tensors.count i32              = 2115
llama_model_loader: - type  f32:  257 tensors
llama_model_loader: - type  f16:   64 tensors
llama_model_loader: - type q8_0:  128 tensors
llama_model_loader: - type q5_K:   64 tensors
llama_model_loader: - type q6_K:    1 tensors
llama_model_loader: - type iq3_xxs:  880 tensors
llama_model_loader: - type iq3_s:  721 tensors

llm_load_print_meta: arch             = grok
llm_load_print_meta: vocab type       = SPM
llm_load_print_meta: n_vocab          = 131072
llm_load_print_meta: n_merges         = 0
llm_load_print_meta: n_ctx_train      = 8192
llm_load_print_meta: n_embd           = 6144
llm_load_print_meta: n_head           = 48
llm_load_print_meta: n_head_kv        = 8
llm_load_print_meta: n_layer          = 64
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            = 6
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             = 32768
llm_load_print_meta: n_expert         = 8
llm_load_print_meta: n_expert_used    = 2
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  = 10000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_yarn_orig_ctx  = 8192
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       = 314B
llm_load_print_meta: model ftype      = IQ3_XS - 3.3 bpw
llm_load_print_meta: model params     = 316.49 B
llm_load_print_meta: model size       = 120.73 GiB (3.28 BPW) 
llm_load_print_meta: general.name     = Grok
llm_load_print_meta: BOS token        = 1 '[BOS]'
llm_load_print_meta: EOS token        = 2 '[EOS]'
llm_load_print_meta: UNK token        = 0 '[PAD]'
llm_load_print_meta: PAD token        = 0 '[PAD]'
llm_load_print_meta: LF token         = 79 '<0x0A>'