Eval bug: Output garbled on DeepSeek-R1-Distill-Qwen-7B-Q4_K_M.gguf from unsloth using musa backend with VMM off #13788
Description
Name and Version
root@f7cd9f1a2456:/ws# ./build/bin/llama-cli --version
ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 MUSA devices:
Device 0: MTT S80, compute capability 2.1, VMM: no
version: 5488 (e121edc)
built with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu
Operating systems
Linux
GGML backends
Musa
Hardware
12th Gen Intel(R) Core(TM) i5-12400 + MTT S80
Models
DeepSeek-R1-Distill-Qwen-7B-Q4_K_M.gguf from unsloth
Problem description & steps to reproduce
git reset 2d77d88e70d017cd82c3f1a4517e3102e2028ac4 --hard- apply diff and build
diff --git a/ggml/src/ggml-musa/CMakeLists.txt b/ggml/src/ggml-musa/CMakeLists.txt
index 92f05d555..eb80418b2 100644
--- a/ggml/src/ggml-musa/CMakeLists.txt
+++ b/ggml/src/ggml-musa/CMakeLists.txt
@@ -75,9 +75,9 @@ if (MUSAToolkit_FOUND)
add_compile_definitions(GGML_CUDA_FORCE_CUBLAS)
endif()
- if (GGML_CUDA_NO_VMM)
+ # if (GGML_CUDA_NO_VMM)
add_compile_definitions(GGML_CUDA_NO_VMM)
- endif()
+ # endif()
if (NOT GGML_CUDA_FA)
add_compile_definitions(GGML_CUDA_NO_FA)- run DeepSeek-R1-Distill-Qwen-7B-Q4_K_M.gguf from unsloth
root@f7cd9f1a2456:/ws# ./build/bin/llama-cli -m /models/DeepSeek-R1-Distill-Qwen-7B-Q4_K_M.gguf -ngl 999
ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 MUSA devices:
Device 0: MTT S80, compute capability 2.1, VMM: no
build: 4953 (2d77d88e7) with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu
main: llama backend init
main: load the model and apply lora adapter, if any
llama_model_load_from_file_impl: using device MUSA0 (MTT S80) - 15752 MiB free
llama_model_loader: loaded meta data with 27 key-value pairs and 339 tensors from /models/DeepSeek-R1-Distill-Qwen-7B-Q4_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.type str = model
llama_model_loader: - kv 2: general.name str = DeepSeek R1 Distill Qwen 7B
llama_model_loader: - kv 3: general.organization str = Deepseek Ai
llama_model_loader: - kv 4: general.basename str = DeepSeek-R1-Distill-Qwen
llama_model_loader: - kv 5: general.size_label str = 7B
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 = 10000.000000
llama_model_loader: - kv 13: qwen2.attention.layer_norm_rms_epsilon f32 = 0.000001
llama_model_loader: - kv 14: tokenizer.ggml.model str = gpt2
llama_model_loader: - kv 15: tokenizer.ggml.pre str = deepseek-r1-qwen
llama_model_loader: - kv 16: tokenizer.ggml.tokens arr[str,152064] = ["!", "\"", "#", "$", "%", "&", "'", ...
llama_model_loader: - kv 17: tokenizer.ggml.token_type arr[i32,152064] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv 18: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
llama_model_loader: - kv 19: tokenizer.ggml.bos_token_id u32 = 151646
llama_model_loader: - kv 20: tokenizer.ggml.eos_token_id u32 = 151643
llama_model_loader: - kv 21: tokenizer.ggml.padding_token_id u32 = 151654
llama_model_loader: - kv 22: tokenizer.ggml.add_bos_token bool = true
llama_model_loader: - kv 23: tokenizer.ggml.add_eos_token bool = false
llama_model_loader: - kv 24: tokenizer.chat_template str = {% if not add_generation_prompt is de...
llama_model_loader: - kv 25: general.quantization_version u32 = 2
llama_model_loader: - kv 26: general.file_type u32 = 15
llama_model_loader: - type f32: 141 tensors
llama_model_loader: - type q4_K: 169 tensors
llama_model_loader: - type q6_K: 29 tensors
print_info: file format = GGUF V3 (latest)
print_info: file type = Q4_K - Medium
print_info: file size = 4.36 GiB (4.91 BPW)
load: special_eos_id is not in special_eog_ids - the tokenizer config may be incorrect
load: special tokens cache size = 22
load: token to piece cache size = 0.9310 MB
print_info: arch = qwen2
print_info: vocab_only = 0
print_info: n_ctx_train = 131072
print_info: n_embd = 3584
print_info: n_layer = 28
print_info: n_head = 28
print_info: n_head_kv = 4
print_info: n_rot = 128
print_info: n_swa = 0
print_info: n_swa_pattern = 1
print_info: n_embd_head_k = 128
print_info: n_embd_head_v = 128
print_info: n_gqa = 7
print_info: n_embd_k_gqa = 512
print_info: n_embd_v_gqa = 512
print_info: f_norm_eps = 0.0e+00
print_info: f_norm_rms_eps = 1.0e-06
print_info: f_clamp_kqv = 0.0e+00
print_info: f_max_alibi_bias = 0.0e+00
print_info: f_logit_scale = 0.0e+00
print_info: f_attn_scale = 0.0e+00
print_info: n_ff = 18944
print_info: n_expert = 0
print_info: n_expert_used = 0
print_info: causal attn = 1
print_info: pooling type = 0
print_info: rope type = 2
print_info: rope scaling = linear
print_info: freq_base_train = 10000.0
print_info: freq_scale_train = 1
print_info: n_ctx_orig_yarn = 131072
print_info: rope_finetuned = unknown
print_info: ssm_d_conv = 0
print_info: ssm_d_inner = 0
print_info: ssm_d_state = 0
print_info: ssm_dt_rank = 0
print_info: ssm_dt_b_c_rms = 0
print_info: model type = 7B
print_info: model params = 7.62 B
print_info: general.name = DeepSeek R1 Distill Qwen 7B
print_info: vocab type = BPE
print_info: n_vocab = 152064
print_info: n_merges = 151387
print_info: BOS token = 151646 '<｜begin▁of▁sentence｜>'
print_info: EOS token = 151643 '<｜end▁of▁sentence｜>'
print_info: EOT token = 151643 '<｜end▁of▁sentence｜>'
print_info: PAD token = 151654 '<|vision_pad|>'
print_info: LF token = 198 'Ċ'
print_info: FIM PRE token = 151659 '<|fim_prefix|>'
print_info: FIM SUF token = 151661 '<|fim_suffix|>'
print_info: FIM MID token = 151660 '<|fim_middle|>'
print_info: FIM PAD token = 151662 '<|fim_pad|>'
print_info: FIM REP token = 151663 '<|repo_name|>'
print_info: FIM SEP token = 151664 '<|file_sep|>'
print_info: EOG token = 151643 '<｜end▁of▁sentence｜>'
print_info: EOG token = 151662 '<|fim_pad|>'
print_info: EOG token = 151663 '<|repo_name|>'
print_info: EOG token = 151664 '<|file_sep|>'
print_info: max token length = 256
load_tensors: loading model tensors, this can take a while... (mmap = true)
make_cpu_buft_list: disabling extra buffer types (i.e. repacking) since a GPU device is available
load_tensors: offloading 28 repeating layers to GPU
load_tensors: offloading output layer to GPU
load_tensors: offloaded 29/29 layers to GPU
load_tensors: MUSA0 model buffer size = 4168.09 MiB
load_tensors: CPU_Mapped model buffer size = 292.36 MiB
..................................................................................
llama_context: constructing llama_context
llama_context: n_seq_max = 1
llama_context: n_ctx = 4096
llama_context: n_ctx_per_seq = 4096
llama_context: n_batch = 2048
llama_context: n_ubatch = 512
llama_context: causal_attn = 1
llama_context: flash_attn = 0
llama_context: freq_base = 10000.0
llama_context: freq_scale = 1
llama_context: n_ctx_per_seq (4096) < n_ctx_train (131072) -- the full capacity of the model will not be utilized
llama_context: MUSA_Host output buffer size = 0.58 MiB
init: kv_size = 4096, offload = 1, type_k = 'f16', type_v = 'f16', n_layer = 28, can_shift = 1
init: MUSA0 KV buffer size = 224.00 MiB
llama_context: KV self size = 224.00 MiB, K (f16): 112.00 MiB, V (f16): 112.00 MiB
llama_context: MUSA0 compute buffer size = 304.00 MiB
llama_context: MUSA_Host compute buffer size = 15.01 MiB
llama_context: graph nodes = 1042
llama_context: graph splits = 2
common_init_from_params: setting dry_penalty_last_n to ctx_size = 4096
common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable)
main: llama threadpool init, n_threads = 6
main: chat template is available, enabling conversation mode (disable it with -no-cnv)
main: chat template example:
You are a helpful assistant
<｜User｜>Hello<｜Assistant｜>Hi there<｜end▁of▁sentence｜><｜User｜>How are you?<｜Assistant｜>
system_info: n_threads = 6 (n_threads_batch = 6) / 12 | MUSA : NO_VMM = 1 | PEER_MAX_BATCH_SIZE = 128 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX_VNNI = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | LLAMAFILE = 1 | OPENMP = 1 | AARCH64_REPACK = 1 |
main: interactive mode on.
sampler seed: 597688009
sampler params:
repeat_last_n = 64, repeat_penalty = 1.000, frequency_penalty = 0.000, presence_penalty = 0.000
dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = 4096
top_k = 40, top_p = 0.950, min_p = 0.050, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 0.800
mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000
sampler chain: logits -> logit-bias -> penalties -> dry -> top-k -> typical -> top-p -> min-p -> xtc -> temp-ext -> dist
generate: n_ctx = 4096, n_batch = 2048, n_predict = -1, n_keep = 1
== Running in interactive mode. ==
- Press Ctrl+C to interject at any time.
- Press Return to return control to the AI.
- To return control without starting a new line, end your input with '/'.
- If you want to submit another line, end your input with '\'.
- Not using system message. To change it, set a different value via -sys PROMPT
> Hi there
UINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINTUINT
> - with
--fa, everything goes fine - mannully revert 2d77d88 on master (e121edc), with or without
-fa, DeepSeek-R1-Distill-Qwen-7B-Q4_K_M.gguf can generate token correctly; other models (nvidia-llama-3_1-nemotron-nano-8b-v1-q4_k_m.gguf, qwen3_8b_q4_k_m.gguf) seems not have this issue; turn off kvcache by using--no-kv-offloadthe issue is gone. I also noticed pp performance downgrade if kvcache is on && disable flash attention (default) - also tried CPU backend and no such issue found
First Bad Commit
Relevant log output
root@f7cd9f1a2456:/ws# ./build/bin/llama-cli -m /models/nvidia-llama-3_1-nemotron-nano-8b-v1-q4_k_m.gguf -ngl 999
ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 MUSA devices:
Device 0: MTT S80, compute capability 2.1, VMM: no
build: 5488 (e121edc43) with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu
main: llama backend init
main: load the model and apply lora adapter, if any
llama_model_load_from_file_impl: using device MUSA0 (MTT S80) - 15752 MiB free
llama_model_loader: loaded meta data with 35 key-value pairs and 292 tensors from /models/nvidia-llama-3_1-nemotron-nano-8b-v1-q4_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 = llama
llama_model_loader: - kv 1: general.type str = model
llama_model_loader: - kv 2: general.name str = Llama 3.1 Nemotron Nano 8B v1
llama_model_loader: - kv 3: general.version str = v1
llama_model_loader: - kv 4: general.organization str = Nvidia
llama_model_loader: - kv 5: general.finetune str = 42f62a403ee352e019834442673256e3fe3de275
llama_model_loader: - kv 6: general.basename str = Llama-3.1-Nemotron-Nano
llama_model_loader: - kv 7: general.size_label str = 8B
llama_model_loader: - kv 8: general.license str = other
llama_model_loader: - kv 9: general.license.name str = nvidia-open-model-license
llama_model_loader: - kv 10: general.license.link str = https://www.nvidia.com/en-us/agreemen...
llama_model_loader: - kv 11: general.tags arr[str,4] = ["nvidia", "llama-3", "pytorch", "tex...
llama_model_loader: - kv 12: general.languages arr[str,1] = ["en"]
llama_model_loader: - kv 13: llama.block_count u32 = 32
llama_model_loader: - kv 14: llama.context_length u32 = 131072
llama_model_loader: - kv 15: llama.embedding_length u32 = 4096
llama_model_loader: - kv 16: llama.feed_forward_length u32 = 14336
llama_model_loader: - kv 17: llama.attention.head_count u32 = 32
llama_model_loader: - kv 18: llama.attention.head_count_kv u32 = 8
llama_model_loader: - kv 19: llama.rope.freq_base f32 = 500000.000000
llama_model_loader: - kv 20: llama.attention.layer_norm_rms_epsilon f32 = 0.000010
llama_model_loader: - kv 21: llama.attention.key_length u32 = 128
llama_model_loader: - kv 22: llama.attention.value_length u32 = 128
llama_model_loader: - kv 23: llama.vocab_size u32 = 128256
llama_model_loader: - kv 24: llama.rope.dimension_count u32 = 128
llama_model_loader: - kv 25: tokenizer.ggml.model str = gpt2
llama_model_loader: - kv 26: tokenizer.ggml.pre str = llama-bpe
llama_model_loader: - kv 27: tokenizer.ggml.tokens arr[str,128256] = ["!", "\"", "#", "$", "%", "&", "'", ...
llama_model_loader: - kv 28: tokenizer.ggml.token_type arr[i32,128256] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv 29: tokenizer.ggml.merges arr[str,280147] = ["Ġ Ġ", "Ġ ĠĠĠ", "ĠĠ ĠĠ", "...
llama_model_loader: - kv 30: tokenizer.ggml.bos_token_id u32 = 128000
llama_model_loader: - kv 31: tokenizer.ggml.eos_token_id u32 = 128009
llama_model_loader: - kv 32: tokenizer.chat_template str = {%- if messages[0]['role'] == 'system...
llama_model_loader: - kv 33: general.quantization_version u32 = 2
llama_model_loader: - kv 34: general.file_type u32 = 15
llama_model_loader: - type f32: 66 tensors
llama_model_loader: - type q4_K: 193 tensors
llama_model_loader: - type q6_K: 33 tensors
print_info: file format = GGUF V3 (latest)
print_info: file type = Q4_K - Medium
print_info: file size = 4.58 GiB (4.89 BPW)
load: special tokens cache size = 256
load: token to piece cache size = 0.7999 MB
print_info: arch = llama
print_info: vocab_only = 0
print_info: n_ctx_train = 131072
print_info: n_embd = 4096
print_info: n_layer = 32
print_info: n_head = 32
print_info: n_head_kv = 8
print_info: n_rot = 128
print_info: n_swa = 0
print_info: is_swa_any = 0
print_info: n_embd_head_k = 128
print_info: n_embd_head_v = 128
print_info: n_gqa = 4
print_info: n_embd_k_gqa = 1024
print_info: n_embd_v_gqa = 1024
print_info: f_norm_eps = 0.0e+00
print_info: f_norm_rms_eps = 1.0e-05
print_info: f_clamp_kqv = 0.0e+00
print_info: f_max_alibi_bias = 0.0e+00
print_info: f_logit_scale = 0.0e+00
print_info: f_attn_scale = 0.0e+00
print_info: n_ff = 14336
print_info: n_expert = 0
print_info: n_expert_used = 0
print_info: causal attn = 1
print_info: pooling type = 0
print_info: rope type = 0
print_info: rope scaling = linear
print_info: freq_base_train = 500000.0
print_info: freq_scale_train = 1
print_info: n_ctx_orig_yarn = 131072
print_info: rope_finetuned = unknown
print_info: ssm_d_conv = 0
print_info: ssm_d_inner = 0
print_info: ssm_d_state = 0
print_info: ssm_dt_rank = 0
print_info: ssm_dt_b_c_rms = 0
print_info: model type = 8B
print_info: model params = 8.03 B
print_info: general.name = Llama 3.1 Nemotron Nano 8B v1
print_info: vocab type = BPE
print_info: n_vocab = 128256
print_info: n_merges = 280147
print_info: BOS token = 128000 '<|begin_of_text|>'
print_info: EOS token = 128009 '<|eot_id|>'
print_info: EOT token = 128009 '<|eot_id|>'
print_info: EOM token = 128008 '<|eom_id|>'
print_info: LF token = 198 'Ċ'
print_info: EOG token = 128008 '<|eom_id|>'
print_info: EOG token = 128009 '<|eot_id|>'
print_info: max token length = 256
load_tensors: loading model tensors, this can take a while... (mmap = true)
load_tensors: offloading 32 repeating layers to GPU
load_tensors: offloading output layer to GPU
load_tensors: offloaded 33/33 layers to GPU
load_tensors: MUSA0 model buffer size = 4403.49 MiB
load_tensors: CPU_Mapped model buffer size = 281.81 MiB
.......................................................................................
llama_context: constructing llama_context
llama_context: n_seq_max = 1
llama_context: n_ctx = 4096
llama_context: n_ctx_per_seq = 4096
llama_context: n_batch = 2048
llama_context: n_ubatch = 512
llama_context: causal_attn = 1
llama_context: flash_attn = 0
llama_context: freq_base = 500000.0
llama_context: freq_scale = 1
llama_context: n_ctx_per_seq (4096) < n_ctx_train (131072) -- the full capacity of the model will not be utilized
llama_context: MUSA_Host output buffer size = 0.49 MiB
llama_kv_cache_unified: MUSA0 KV buffer size = 512.00 MiB
llama_kv_cache_unified: size = 512.00 MiB ( 4096 cells, 32 layers, 1 seqs), K (f16): 256.00 MiB, V (f16): 256.00 MiB
llama_context: MUSA0 compute buffer size = 296.00 MiB
llama_context: MUSA_Host compute buffer size = 16.01 MiB
llama_context: graph nodes = 1158
llama_context: graph splits = 2
common_init_from_params: setting dry_penalty_last_n to ctx_size = 4096
common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable)
main: llama threadpool init, n_threads = 6
main: chat template is available, enabling conversation mode (disable it with -no-cnv)
main: chat template example:
<|start_header_id|>system<|end_header_id|>
You are a helpful assistant<|eot_id|><|start_header_id|>user<|end_header_id|>
Hello<|eot_id|><|start_header_id|>assistant<|end_header_id|>
Hi there<|eot_id|><|start_header_id|>user<|end_header_id|>
How are you?<|eot_id|><|start_header_id|>assistant<|end_header_id|>
system_info: n_threads = 6 (n_threads_batch = 6) / 12 | MUSA : NO_VMM = 1 | PEER_MAX_BATCH_SIZE = 128 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX_VNNI = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | LLAMAFILE = 1 | OPENMP = 1 | AARCH64_REPACK = 1 |
main: interactive mode on.
sampler seed: 3669391543
sampler params:
repeat_last_n = 64, repeat_penalty = 1.000, frequency_penalty = 0.000, presence_penalty = 0.000
dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = 4096
top_k = 40, top_p = 0.950, min_p = 0.050, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 0.800
mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000
sampler chain: logits -> logit-bias -> penalties -> dry -> top-n-sigma -> top-k -> typical -> top-p -> min-p -> xtc -> temp-ext -> dist
generate: n_ctx = 4096, n_batch = 2048, n_predict = -1, n_keep = 1
== Running in interactive mode. ==
- Press Ctrl+C to interject at any time.
- Press Return to return control to the AI.
- To return control without starting a new line, end your input with '/'.
- If you want to submit another line, end your input with '\'.
- Not using system message. To change it, set a different value via -sys PROMPT
> Hi
Hello! It's nice to meet you. How can I assist you today?
>
llama_perf_sampler_print: sampling time = 1.48 ms / 27 runs ( 0.05 ms per token, 18218.62 tokens per second)
llama_perf_context_print: load time = 2201.14 ms
llama_perf_context_print: prompt eval time = 4094.29 ms / 11 tokens ( 372.21 ms per token, 2.69 tokens per second)
llama_perf_context_print: eval time = 1359.07 ms / 16 runs ( 84.94 ms per token, 11.77 tokens per second)
llama_perf_context_print: total time = 8089.12 ms / 27 tokens
Interrupted by user
root@f7cd9f1a2456:/ws# ./build/bin/llama-cli -m /models/nvidia-llama-3_1-nemotron-nano-8b-v1-q4_k_m.gguf -ngl 999 -fa
ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 MUSA devices:
Device 0: MTT S80, compute capability 2.1, VMM: no
build: 5488 (e121edc43) with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu
main: llama backend init
main: load the model and apply lora adapter, if any
llama_model_load_from_file_impl: using device MUSA0 (MTT S80) - 15752 MiB free
llama_model_loader: loaded meta data with 35 key-value pairs and 292 tensors from /models/nvidia-llama-3_1-nemotron-nano-8b-v1-q4_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 = llama
llama_model_loader: - kv 1: general.type str = model
llama_model_loader: - kv 2: general.name str = Llama 3.1 Nemotron Nano 8B v1
llama_model_loader: - kv 3: general.version str = v1
llama_model_loader: - kv 4: general.organization str = Nvidia
llama_model_loader: - kv 5: general.finetune str = 42f62a403ee352e019834442673256e3fe3de275
llama_model_loader: - kv 6: general.basename str = Llama-3.1-Nemotron-Nano
llama_model_loader: - kv 7: general.size_label str = 8B
llama_model_loader: - kv 8: general.license str = other
llama_model_loader: - kv 9: general.license.name str = nvidia-open-model-license
llama_model_loader: - kv 10: general.license.link str = https://www.nvidia.com/en-us/agreemen...
llama_model_loader: - kv 11: general.tags arr[str,4] = ["nvidia", "llama-3", "pytorch", "tex...
llama_model_loader: - kv 12: general.languages arr[str,1] = ["en"]
llama_model_loader: - kv 13: llama.block_count u32 = 32
llama_model_loader: - kv 14: llama.context_length u32 = 131072
llama_model_loader: - kv 15: llama.embedding_length u32 = 4096
llama_model_loader: - kv 16: llama.feed_forward_length u32 = 14336
llama_model_loader: - kv 17: llama.attention.head_count u32 = 32
llama_model_loader: - kv 18: llama.attention.head_count_kv u32 = 8
llama_model_loader: - kv 19: llama.rope.freq_base f32 = 500000.000000
llama_model_loader: - kv 20: llama.attention.layer_norm_rms_epsilon f32 = 0.000010
llama_model_loader: - kv 21: llama.attention.key_length u32 = 128
llama_model_loader: - kv 22: llama.attention.value_length u32 = 128
llama_model_loader: - kv 23: llama.vocab_size u32 = 128256
llama_model_loader: - kv 24: llama.rope.dimension_count u32 = 128
llama_model_loader: - kv 25: tokenizer.ggml.model str = gpt2
llama_model_loader: - kv 26: tokenizer.ggml.pre str = llama-bpe
llama_model_loader: - kv 27: tokenizer.ggml.tokens arr[str,128256] = ["!", "\"", "#", "$", "%", "&", "'", ...
llama_model_loader: - kv 28: tokenizer.ggml.token_type arr[i32,128256] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv 29: tokenizer.ggml.merges arr[str,280147] = ["Ġ Ġ", "Ġ ĠĠĠ", "ĠĠ ĠĠ", "...
llama_model_loader: - kv 30: tokenizer.ggml.bos_token_id u32 = 128000
llama_model_loader: - kv 31: tokenizer.ggml.eos_token_id u32 = 128009
llama_model_loader: - kv 32: tokenizer.chat_template str = {%- if messages[0]['role'] == 'system...
llama_model_loader: - kv 33: general.quantization_version u32 = 2
llama_model_loader: - kv 34: general.file_type u32 = 15
llama_model_loader: - type f32: 66 tensors
llama_model_loader: - type q4_K: 193 tensors
llama_model_loader: - type q6_K: 33 tensors
print_info: file format = GGUF V3 (latest)
print_info: file type = Q4_K - Medium
print_info: file size = 4.58 GiB (4.89 BPW)
load: special tokens cache size = 256
load: token to piece cache size = 0.7999 MB
print_info: arch = llama
print_info: vocab_only = 0
print_info: n_ctx_train = 131072
print_info: n_embd = 4096
print_info: n_layer = 32
print_info: n_head = 32
print_info: n_head_kv = 8
print_info: n_rot = 128
print_info: n_swa = 0
print_info: is_swa_any = 0
print_info: n_embd_head_k = 128
print_info: n_embd_head_v = 128
print_info: n_gqa = 4
print_info: n_embd_k_gqa = 1024
print_info: n_embd_v_gqa = 1024
print_info: f_norm_eps = 0.0e+00
print_info: f_norm_rms_eps = 1.0e-05
print_info: f_clamp_kqv = 0.0e+00
print_info: f_max_alibi_bias = 0.0e+00
print_info: f_logit_scale = 0.0e+00
print_info: f_attn_scale = 0.0e+00
print_info: n_ff = 14336
print_info: n_expert = 0
print_info: n_expert_used = 0
print_info: causal attn = 1
print_info: pooling type = 0
print_info: rope type = 0
print_info: rope scaling = linear
print_info: freq_base_train = 500000.0
print_info: freq_scale_train = 1
print_info: n_ctx_orig_yarn = 131072
print_info: rope_finetuned = unknown
print_info: ssm_d_conv = 0
print_info: ssm_d_inner = 0
print_info: ssm_d_state = 0
print_info: ssm_dt_rank = 0
print_info: ssm_dt_b_c_rms = 0
print_info: model type = 8B
print_info: model params = 8.03 B
print_info: general.name = Llama 3.1 Nemotron Nano 8B v1
print_info: vocab type = BPE
print_info: n_vocab = 128256
print_info: n_merges = 280147
print_info: BOS token = 128000 '<|begin_of_text|>'
print_info: EOS token = 128009 '<|eot_id|>'
print_info: EOT token = 128009 '<|eot_id|>'
print_info: EOM token = 128008 '<|eom_id|>'
print_info: LF token = 198 'Ċ'
print_info: EOG token = 128008 '<|eom_id|>'
print_info: EOG token = 128009 '<|eot_id|>'
print_info: max token length = 256
load_tensors: loading model tensors, this can take a while... (mmap = true)
load_tensors: offloading 32 repeating layers to GPU
load_tensors: offloading output layer to GPU
load_tensors: offloaded 33/33 layers to GPU
load_tensors: MUSA0 model buffer size = 4403.49 MiB
load_tensors: CPU_Mapped model buffer size = 281.81 MiB
.......................................................................................
llama_context: constructing llama_context
llama_context: n_seq_max = 1
llama_context: n_ctx = 4096
llama_context: n_ctx_per_seq = 4096
llama_context: n_batch = 2048
llama_context: n_ubatch = 512
llama_context: causal_attn = 1
llama_context: flash_attn = 1
llama_context: freq_base = 500000.0
llama_context: freq_scale = 1
llama_context: n_ctx_per_seq (4096) < n_ctx_train (131072) -- the full capacity of the model will not be utilized
llama_context: MUSA_Host output buffer size = 0.49 MiB
llama_kv_cache_unified: MUSA0 KV buffer size = 512.00 MiB
llama_kv_cache_unified: size = 512.00 MiB ( 4096 cells, 32 layers, 1 seqs), K (f16): 256.00 MiB, V (f16): 256.00 MiB
llama_context: MUSA0 compute buffer size = 266.55 MiB
llama_context: MUSA_Host compute buffer size = 36.01 MiB
llama_context: graph nodes = 1031
llama_context: graph splits = 66
common_init_from_params: setting dry_penalty_last_n to ctx_size = 4096
common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable)
main: llama threadpool init, n_threads = 6
main: chat template is available, enabling conversation mode (disable it with -no-cnv)
main: chat template example:
<|start_header_id|>system<|end_header_id|>
You are a helpful assistant<|eot_id|><|start_header_id|>user<|end_header_id|>
Hello<|eot_id|><|start_header_id|>assistant<|end_header_id|>
Hi there<|eot_id|><|start_header_id|>user<|end_header_id|>
How are you?<|eot_id|><|start_header_id|>assistant<|end_header_id|>
system_info: n_threads = 6 (n_threads_batch = 6) / 12 | MUSA : NO_VMM = 1 | PEER_MAX_BATCH_SIZE = 128 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX_VNNI = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | LLAMAFILE = 1 | OPENMP = 1 | AARCH64_REPACK = 1 |
main: interactive mode on.
sampler seed: 401977407
sampler params:
repeat_last_n = 64, repeat_penalty = 1.000, frequency_penalty = 0.000, presence_penalty = 0.000
dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = 4096
top_k = 40, top_p = 0.950, min_p = 0.050, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 0.800
mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000
sampler chain: logits -> logit-bias -> penalties -> dry -> top-n-sigma -> top-k -> typical -> top-p -> min-p -> xtc -> temp-ext -> dist
generate: n_ctx = 4096, n_batch = 2048, n_predict = -1, n_keep = 1
== Running in interactive mode. ==
- Press Ctrl+C to interject at any time.
- Press Return to return control to the AI.
- To return control without starting a new line, end your input with '/'.
- If you want to submit another line, end your input with '\'.
- Not using system message. To change it, set a different value via -sys PROMPT
> Hi
Hello! How can I assist you today?
>
llama_perf_sampler_print: sampling time = 0.90 ms / 20 runs ( 0.05 ms per token, 22148.39 tokens per second)
llama_perf_context_print: load time = 944.39 ms
llama_perf_context_print: prompt eval time = 781.78 ms / 11 tokens ( 71.07 ms per token, 14.07 tokens per second)
llama_perf_context_print: eval time = 810.46 ms / 9 runs ( 90.05 ms per token, 11.10 tokens per second)
llama_perf_context_print: total time = 10864.23 ms / 20 tokens
Interrupted by user
root@f7cd9f1a2456:/ws#