tools

47 5 weeks ago

e95f18fa010c · 4.8GB
    Metadata
  • general.architecture
    command-r
  • general.file_type
    IQ4_NL
  • command-r.attention.head_count
    32
  • command-r.attention.head_count_kv
    8
  • command-r.attention.layer_norm_epsilon
    1e-05
  • command-r.block_count
    32
  • command-r.context_length
    8192
  • command-r.embedding_length
    4096
  • command-r.feed_forward_length
    14336
  • command-r.logit_scale
    0.125
  • command-r.rope.freq_base
    10000
  • command-r.rope.scaling.type
    none
  • tokenizer.chat_template.rag
    {{ bos_token }}{% if messages[0]['role'] == 'system' %}{% set loop_messages = messages[1:] %}{% set system_message = messages[0]['content'] %}{% else %}{% set loop_messages = messages %}{% set system_message = '## Task and Context You help people answer their questions and other requests interactively...' %}{% endif %}{{ '<|START_OF_TURN_TOKEN|><|SYSTEM_TOKEN|>' }}{{ '# Safety Preamble' }}{{ ' The instructions in this section override those in the task description and style guide sections. Don't answer questions that are harmful or immoral.' }}{{ ' # System Preamble' }}{{ ' ## Basic Rules' }}{{ ' You are a powerful conversational AI trained by Cohere...' }}{% for message in loop_messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '<|START_OF_TURN_TOKEN|><|USER_TOKEN|>' + content.strip() + '<|END_OF_TURN_TOKEN|>' }}{% elif message['role'] == 'assistant' %}{{ '<|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>' + content.strip() + '<|END_OF_TURN_TOKEN|>' }}{% endif %}{% endfor %}{{ '<|START_OF_TURN_TOKEN|><|SYSTEM_TOKEN|>' }}{{ '<results>' }}{% for document in documents %}{{ ' Document: ' }}{{ loop.index0 }} {% for key, value in document.items() %}{{ key }}: {{ value }} {% endfor %}{% endfor %}{{ '</results>' }}{{ '<|END_OF_TURN_TOKEN|>' }}
  • tokenizer.chat_template.tool_use
    {{ bos_token }}{% if messages[0]['role'] == 'system' %}{% set loop_messages = messages[1:] %}{% set system_message = messages[0]['content'] %}{% else %}{% set loop_messages = messages %}{% set system_message = '## Task and Context You help people answer their questions and other requests interactively...' %}{% endif %}{{ '<|START_OF_TURN_TOKEN|><|SYSTEM_TOKEN|>' }}{{ '# Safety Preamble' }}{{ ' The instructions in this section override those in the task description and style guide sections. Don't answer questions that are harmful or immoral.' }}{{ ' # System Preamble' }}{{ ' ## Basic Rules' }}{{ ' You are a powerful conversational AI trained by Mann-E Labs...' }}{% for tool in tools %}{{'```python def ' + tool.name + '('}}{% for param_name, param_fields in tool.parameter_definitions.items() %}{% if loop.index0 != 0 %}{{ ', ' }}{% endif %}{{ param_name }}: {{ param_fields.type if param_fields.required else 'Optional[' + param_fields.type + '] = None' }}{% endfor %}{{ ') -> List[Dict]: + tool.description + pass ```' }}{% endfor %}
  • tokenizer.chat_templates
    [rag, tool_use]
  • tokenizer.ggml.add_bos_token
    true
  • tokenizer.ggml.add_eos_token
    false
  • tokenizer.ggml.bos_token_id
    5
  • tokenizer.ggml.eos_token_id
    255001
  • tokenizer.ggml.merges
    [Ġ Ġ, Ġ t, e r, i n, Ġ a, ...]
  • tokenizer.ggml.model
    gpt2
  • tokenizer.ggml.padding_token_id
    0
  • tokenizer.ggml.pre
    command-r
  • tokenizer.ggml.token_type
    [3, 3, 3, 3, 3, ...]
  • tokenizer.ggml.tokens
    [<PAD>, <UNK>, <CLS>, <SEP>, <MASK_TOKEN>, ...]
  • quantize.imatrix.chunks_count
    40
  • quantize.imatrix.dataset
    /tmp/gradio/ccd8592fbb2dd6b90cf7e56f254cf126949e2276d35c98b91fe1a6172dae9d86/zhco.txt
  • quantize.imatrix.entries_count
    224
  • quantize.imatrix.file
    outputs/tmpyfuxk4s4/imatrix.dat
  • Tensor
  • token_embd.weight
    Q6_K
    [4096, 256000]
  • blk.0
  • blk.0.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.0.attn_norm.weight
    F32
    [4096]
  • blk.0.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.0.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.0.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.0.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.0.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.0.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.1
  • blk.1.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.1.attn_norm.weight
    F32
    [4096]
  • blk.1.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.1.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.1.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.1.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.1.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.1.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.2
  • blk.2.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.2.attn_norm.weight
    F32
    [4096]
  • blk.2.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.2.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.2.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.2.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.2.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.2.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.3
  • blk.3.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.3.attn_norm.weight
    F32
    [4096]
  • blk.3.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.3.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.3.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.3.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.3.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.3.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.4
  • blk.4.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.4.attn_norm.weight
    F32
    [4096]
  • blk.4.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.4.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.4.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.4.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.4.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.4.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.5
  • blk.5.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.5.attn_norm.weight
    F32
    [4096]
  • blk.5.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.5.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.5.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.5.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.5.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.5.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.6
  • blk.6.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.6.attn_norm.weight
    F32
    [4096]
  • blk.6.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.6.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.6.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.6.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.6.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.6.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.7
  • blk.7.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.7.attn_norm.weight
    F32
    [4096]
  • blk.7.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.7.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.7.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.7.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.7.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.7.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.8
  • blk.8.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.8.attn_norm.weight
    F32
    [4096]
  • blk.8.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.8.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.8.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.8.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.8.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.8.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.9
  • blk.9.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.9.attn_norm.weight
    F32
    [4096]
  • blk.9.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.9.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.9.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.9.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.9.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.9.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.10
  • blk.10.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.10.attn_norm.weight
    F32
    [4096]
  • blk.10.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.10.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.10.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.10.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.10.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.10.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.11
  • blk.11.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.11.attn_norm.weight
    F32
    [4096]
  • blk.11.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.11.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.11.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.11.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.11.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.11.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.12
  • blk.12.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.12.attn_norm.weight
    F32
    [4096]
  • blk.12.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.12.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.12.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.12.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.12.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.12.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.13
  • blk.13.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.13.attn_norm.weight
    F32
    [4096]
  • blk.13.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.13.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.13.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.13.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.13.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.13.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.14
  • blk.14.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.14.attn_norm.weight
    F32
    [4096]
  • blk.14.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.14.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.14.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.14.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.14.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.14.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.15
  • blk.15.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.15.attn_norm.weight
    F32
    [4096]
  • blk.15.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.15.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.15.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.15.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.15.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.15.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.16
  • blk.16.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.16.attn_norm.weight
    F32
    [4096]
  • blk.16.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.16.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.16.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.16.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.16.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.16.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.17
  • blk.17.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.17.attn_norm.weight
    F32
    [4096]
  • blk.17.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.17.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.17.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.17.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.17.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.17.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.18
  • blk.18.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.18.attn_norm.weight
    F32
    [4096]
  • blk.18.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.18.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.18.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.18.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.18.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.18.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.19
  • blk.19.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.19.attn_norm.weight
    F32
    [4096]
  • blk.19.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.19.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.19.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.19.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.19.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.19.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.20
  • blk.20.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.20.attn_norm.weight
    F32
    [4096]
  • blk.20.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.20.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.20.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.20.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.20.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.20.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.21
  • blk.21.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.21.attn_norm.weight
    F32
    [4096]
  • blk.21.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.21.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.21.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.21.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.21.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.21.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.22
  • blk.22.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.22.attn_norm.weight
    F32
    [4096]
  • blk.22.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.22.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.22.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.22.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.22.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.22.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.23
  • blk.23.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.23.attn_norm.weight
    F32
    [4096]
  • blk.23.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.23.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.23.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.23.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.23.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.23.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.24
  • blk.24.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.24.attn_norm.weight
    F32
    [4096]
  • blk.24.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.24.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.24.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.24.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.24.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.24.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.25
  • blk.25.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.25.attn_norm.weight
    F32
    [4096]
  • blk.25.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.25.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.25.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.25.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.25.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.25.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.26
  • blk.26.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.26.attn_norm.weight
    F32
    [4096]
  • blk.26.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.26.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.26.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.26.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.26.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.26.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.27
  • blk.27.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.27.attn_norm.weight
    F32
    [4096]
  • blk.27.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.27.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.27.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.27.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.27.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.27.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.28
  • blk.28.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.28.attn_norm.weight
    F32
    [4096]
  • blk.28.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.28.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.28.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.28.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.28.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.28.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.29
  • blk.29.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.29.attn_norm.weight
    F32
    [4096]
  • blk.29.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.29.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.29.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.29.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.29.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.29.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.30
  • blk.30.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.30.attn_norm.weight
    F32
    [4096]
  • blk.30.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.30.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.30.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.30.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.30.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.30.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • blk.31
  • blk.31.attn_k.weight
    IQ4_NL
    [4096, 1024]
  • blk.31.attn_norm.weight
    F32
    [4096]
  • blk.31.attn_output.weight
    IQ4_NL
    [4096, 4096]
  • blk.31.attn_q.weight
    IQ4_NL
    [4096, 4096]
  • blk.31.attn_v.weight
    Q5_K
    [4096, 1024]
  • blk.31.ffn_down.weight
    IQ4_NL
    [14336, 4096]
  • blk.31.ffn_gate.weight
    IQ4_NL
    [4096, 14336]
  • blk.31.ffn_up.weight
    IQ4_NL
    [4096, 14336]
  • output_norm.weight
    F32
    [4096]