glaiveai/glaive-function-calling-v2-small

1"""Attention layers."""2import math3import warnings4from typing import Optional5import torch6import torch.nn as nn7from einops import rearrange8from torch import nn9from .norm import LPLayerNorm10 11def _reset_is_causal(num_query_tokens: int, num_key_tokens: int, original_is_causal: bool):12    if original_is_causal and num_query_tokens != num_key_tokens:13        if num_query_tokens != 1:14            raise NotImplementedError('MPT does not support query and key with different number of tokens, unless number of query tokens is 1.')15        else:16            return False17    return original_is_causal18 19def scaled_multihead_dot_product_attention(query, key, value, n_heads, softmax_scale=None, attn_bias=None, key_padding_mask=None, is_causal=False, dropout_p=0.0, training=False, needs_weights=False, multiquery=False):20    q = rearrange(query, 'b s (h d) -> b h s d', h=n_heads)21    k = rearrange(key, 'b s (h d) -> b h d s', h=1 if multiquery else n_heads)22    v = rearrange(value, 'b s (h d) -> b h s d', h=1 if multiquery else n_heads)23    min_val = torch.finfo(q.dtype).min24    (b, _, s_q, d) = q.shape25    s_k = k.size(-1)26    if softmax_scale is None:27        softmax_scale = 1 / math.sqrt(d)28    attn_weight = q.matmul(k) * softmax_scale29    if attn_bias is not None:30        if attn_bias.size(-1) != 1 and attn_bias.size(-1) != s_k or (attn_bias.size(-2) != 1 and attn_bias.size(-2) != s_q):31            raise RuntimeError(f'attn_bias (shape: {attn_bias.shape}) is expected to broadcast to shape: {attn_weight.shape}.')32        attn_weight = attn_weight + attn_bias33    if key_padding_mask is not None:34        if attn_bias is not None:35            warnings.warn('Propogating key_padding_mask to the attention module ' + 'and applying it within the attention module can cause ' + 'unneccessary computation/memory usage. Consider integrating ' + 'into attn_bias once and passing that to each attention ' + 'module instead.')36        attn_weight = attn_weight.masked_fill(~key_padding_mask.view((b, 1, 1, s_k)), min_val)37    if is_causal:38        s = max(s_q, s_k)39        causal_mask = attn_weight.new_ones(s, s, dtype=torch.float16)40        causal_mask = causal_mask.tril()41        causal_mask = causal_mask.to(torch.bool)42        causal_mask = ~causal_mask43        causal_mask = causal_mask[-s_q:, -s_k:]44        attn_weight = attn_weight.masked_fill(causal_mask.view(1, 1, s_q, s_k), min_val)45    attn_weight = torch.softmax(attn_weight, dim=-1)46    if dropout_p:47        attn_weight = torch.nn.functional.dropout(attn_weight, p=dropout_p, training=training, inplace=True)48    out = attn_weight.matmul(v)49    out = rearrange(out, 'b h s d -> b s (h d)')50    if needs_weights:51        return (out, attn_weight)52    return (out, None)53 54def check_valid_inputs(*tensors, valid_dtypes=[torch.float16, torch.bfloat16]):55    for tensor in tensors:56        if tensor.dtype not in valid_dtypes:57            raise TypeError(f'tensor.dtype={tensor.dtype!r} must be in valid_dtypes={valid_dtypes!r}.')58        if not tensor.is_cuda:59            raise TypeError(f'Inputs must be cuda tensors (tensor.is_cuda={tensor.is_cuda!r}).')60 61def flash_attn_fn(query, key, value, n_heads, softmax_scale=None, attn_bias=None, key_padding_mask=None, is_causal=False, dropout_p=0.0, training=False, needs_weights=False, multiquery=False):62    try:63        from flash_attn import bert_padding, flash_attn_interface64    except:65        raise RuntimeError('Please install flash-attn==1.0.3.post0')66    check_valid_inputs(query, key, value)67    if attn_bias is not None:68        raise NotImplementedError(f'attn_bias not implemented for flash attn.')69    (batch_size, seqlen) = query.shape[:2]70    if key_padding_mask is None:71        key_padding_mask = torch.ones_like(key[:, :, 0], dtype=torch.bool)72    query_padding_mask = key_padding_mask[:, -query.size(1):]73    (query_unpad, indices_q, cu_seqlens_q, max_seqlen_q) = bert_padding.unpad_input(query, query_padding_mask)74    query_unpad = rearrange(query_unpad, 'nnz (h d) -> nnz h d', h=n_heads)75    (key_unpad, _, cu_seqlens_k, max_seqlen_k) = bert_padding.unpad_input(key, key_padding_mask)76    key_unpad = rearrange(key_unpad, 'nnz (h d) -> nnz h d', h=1 if multiquery else n_heads)77    (value_unpad, _, _, _) = bert_padding.unpad_input(value, key_padding_mask)78    value_unpad = rearrange(value_unpad, 'nnz (h d) -> nnz h d', h=1 if multiquery else n_heads)79    if multiquery:80        key_unpad = key_unpad.expand(key_unpad.size(0), n_heads, key_unpad.size(-1))81        value_unpad = value_unpad.expand(value_unpad.size(0), n_heads, value_unpad.size(-1))82    dropout_p = dropout_p if training else 0.083    reset_is_causal = _reset_is_causal(query.size(1), key.size(1), is_causal)84    output_unpad = flash_attn_interface.flash_attn_unpadded_func(query_unpad, key_unpad, value_unpad, cu_seqlens_q, cu_seqlens_k, max_seqlen_q, max_seqlen_k, dropout_p, softmax_scale=softmax_scale, causal=reset_is_causal, return_attn_probs=needs_weights)85    output = bert_padding.pad_input(rearrange(output_unpad, 'nnz h d -> nnz (h d)'), indices_q, batch_size, seqlen)86    return (output, None)87 88def triton_flash_attn_fn(query, key, value, n_heads, softmax_scale=None, attn_bias=None, key_padding_mask=None, is_causal=False, dropout_p=0.0, training=False, needs_weights=False, multiquery=False):89    try:90        from flash_attn import flash_attn_triton91    except:92        raise RuntimeError('Please install flash-attn==1.0.3.post0 and triton==2.0.0.dev20221202')93    check_valid_inputs(query, key, value)94    if dropout_p:95        raise NotImplementedError(f'Dropout not implemented for attn_impl: triton.')96    if needs_weights:97        raise NotImplementedError(f'attn_impl: triton cannot return attn weights.')98    if key_padding_mask is not None:99        warnings.warn('Propagating key_padding_mask to the attention module ' + 'and applying it within the attention module can cause ' + 'unnecessary computation/memory usage. Consider integrating ' + 'into attn_bias once and passing that to each attention ' + 'module instead.')100        (b_size, s_k) = key_padding_mask.shape[:2]101        if attn_bias is None:102            attn_bias = query.new_zeros(b_size, 1, 1, s_k)103        attn_bias = attn_bias.masked_fill(~key_padding_mask.view((b_size, 1, 1, s_k)), torch.finfo(query.dtype).min)104    query = rearrange(query, 'b s (h d) -> b s h d', h=n_heads)105    key = rearrange(key, 'b s (h d) -> b s h d', h=1 if multiquery else n_heads)106    value = rearrange(value, 'b s (h d) -> b s h d', h=1 if multiquery else n_heads)107    if multiquery:108        key = key.expand(*key.shape[:2], n_heads, key.size(-1))109        value = value.expand(*value.shape[:2], n_heads, value.size(-1))110    reset_is_causal = _reset_is_causal(query.size(1), key.size(1), is_causal)111    attn_output = flash_attn_triton.flash_attn_func(query, key, value, attn_bias, reset_is_causal, softmax_scale)112    output = attn_output.view(*attn_output.shape[:2], -1)113    return (output, None)114 115class MultiheadAttention(nn.Module):116    """Multi-head self attention.117 118    Using torch or triton attention implemetation enables user to also use119    additive bias.120    """121 122    def __init__(self, d_model: int, n_heads: int, attn_impl: str='triton', clip_qkv: Optional[float]=None, qk_ln: bool=False, softmax_scale: Optional[float]=None, attn_pdrop: float=0.0, low_precision_layernorm: bool=False, verbose: int=0, device: Optional[str]=None):123        super().__init__()124        self.attn_impl = attn_impl125        self.clip_qkv = clip_qkv126        self.qk_ln = qk_ln127        self.d_model = d_model128        self.n_heads = n_heads129        self.softmax_scale = softmax_scale130        if self.softmax_scale is None:131            self.softmax_scale = 1 / math.sqrt(self.d_model / self.n_heads)132        self.attn_dropout_p = attn_pdrop133        self.Wqkv = nn.Linear(self.d_model, 3 * self.d_model, device=device)134        fuse_splits = (d_model, 2 * d_model)135        self.Wqkv._fused = (0, fuse_splits)136        if self.qk_ln:137            layernorm_class = LPLayerNorm if low_precision_layernorm else nn.LayerNorm138            self.q_ln = layernorm_class(self.d_model, device=device)139            self.k_ln = layernorm_class(self.d_model, device=device)140        if self.attn_impl == 'flash':141            self.attn_fn = flash_attn_fn142        elif self.attn_impl == 'triton':143            self.attn_fn = triton_flash_attn_fn144            if verbose:145                warnings.warn('While `attn_impl: triton` can be faster than `attn_impl: flash` ' + 'it uses more memory. When training larger models this can trigger ' + 'alloc retries which hurts performance. If encountered, we recommend ' + 'using `attn_impl: flash` if your model does not use `alibi` or `prefix_lm`.')146        elif self.attn_impl == 'torch':147            self.attn_fn = scaled_multihead_dot_product_attention148            if torch.cuda.is_available() and verbose:149                warnings.warn('Using `attn_impl: torch`. If your model does not use `alibi` or ' + '`prefix_lm` we recommend using `attn_impl: flash` otherwise ' + 'we recommend using `attn_impl: triton`.')150        else:151            raise ValueError(f'attn_impl={attn_impl!r} is an invalid setting.')152        self.out_proj = nn.Linear(self.d_model, self.d_model, device=device)153        self.out_proj._is_residual = True154 155    def forward(self, x, past_key_value=None, attn_bias=None, attention_mask=None, is_causal=True, needs_weights=False):156        qkv = self.Wqkv(x)157        if self.clip_qkv:158            qkv.clamp_(min=-self.clip_qkv, max=self.clip_qkv)159        (query, key, value) = qkv.chunk(3, dim=2)160        key_padding_mask = attention_mask161        if self.qk_ln:162            dtype = query.dtype163            query = self.q_ln(query).to(dtype)164            key = self.k_ln(key).to(dtype)165        if past_key_value is not None:166            if len(past_key_value) != 0:167                key = torch.cat([past_key_value[0], key], dim=1)168                value = torch.cat([past_key_value[1], value], dim=1)169            past_key_value = (key, value)170        if attn_bias is not None:171            attn_bias = attn_bias[:, :, -query.size(1):, -key.size(1):]172        (context, attn_weights) = self.attn_fn(query, key, value, self.n_heads, softmax_scale=self.softmax_scale, attn_bias=attn_bias, key_padding_mask=key_padding_mask, is_causal=is_causal, dropout_p=self.attn_dropout_p, training=self.training, needs_weights=needs_weights)173        return (self.out_proj(context), attn_weights, past_key_value)174 175class MultiQueryAttention(nn.Module):176    """Multi-Query self attention.177 178    Using torch or triton attention implemetation enables user to also use179    additive bias.180    """181 182    def __init__(self, d_model: int, n_heads: int, attn_impl: str='triton', clip_qkv: Optional[float]=None, qk_ln: bool=False, softmax_scale: Optional[float]=None, attn_pdrop: float=0.0, low_precision_layernorm: bool=False, verbose: int=0, device: Optional[str]=None):183        super().__init__()184        self.attn_impl = attn_impl185        self.clip_qkv = clip_qkv186        self.qk_ln = qk_ln187        self.d_model = d_model188        self.n_heads = n_heads189        self.head_dim = d_model // n_heads190        self.softmax_scale = softmax_scale191        if self.softmax_scale is None:192            self.softmax_scale = 1 / math.sqrt(self.head_dim)193        self.attn_dropout_p = attn_pdrop194        self.Wqkv = nn.Linear(d_model, d_model + 2 * self.head_dim, device=device)195        fuse_splits = (d_model, d_model + self.head_dim)196        self.Wqkv._fused = (0, fuse_splits)197        if self.qk_ln:198            layernorm_class = LPLayerNorm if low_precision_layernorm else nn.LayerNorm199            self.q_ln = layernorm_class(d_model, device=device)200            self.k_ln = layernorm_class(self.head_dim, device=device)201        if self.attn_impl == 'flash':202            self.attn_fn = flash_attn_fn203        elif self.attn_impl == 'triton':204            self.attn_fn = triton_flash_attn_fn205            if verbose:206                warnings.warn('While `attn_impl: triton` can be faster than `attn_impl: flash` ' + 'it uses more memory. When training larger models this can trigger ' + 'alloc retries which hurts performance. If encountered, we recommend ' + 'using `attn_impl: flash` if your model does not use `alibi` or `prefix_lm`.')207        elif self.attn_impl == 'torch':208            self.attn_fn = scaled_multihead_dot_product_attention209            if torch.cuda.is_available() and verbose:210                warnings.warn('Using `attn_impl: torch`. If your model does not use `alibi` or ' + '`prefix_lm` we recommend using `attn_impl: flash` otherwise ' + 'we recommend using `attn_impl: triton`.')211        else:212            raise ValueError(f'attn_impl={attn_impl!r} is an invalid setting.')213        self.out_proj = nn.Linear(self.d_model, self.d_model, device=device)214        self.out_proj._is_residual = True215 216    def forward(self, x, past_key_value=None, attn_bias=None, attention_mask=None, is_causal=True, needs_weights=False):217        qkv = self.Wqkv(x)218        if self.clip_qkv:219            qkv.clamp_(min=-self.clip_qkv, max=self.clip_qkv)220        (query, key, value) = qkv.split([self.d_model, self.head_dim, self.head_dim], dim=2)221        key_padding_mask = attention_mask222        if self.qk_ln:223            dtype = query.dtype224            query = self.q_ln(query).to(dtype)225            key = self.k_ln(key).to(dtype)226        if past_key_value is not None:227            if len(past_key_value) != 0:228                key = torch.cat([past_key_value[0], key], dim=1)229                value = torch.cat([past_key_value[1], value], dim=1)230            past_key_value = (key, value)231        if attn_bias is not None:232            attn_bias = attn_bias[:, :, -query.size(1):, -key.size(1):]233        (context, attn_weights) = self.attn_fn(query, key, value, self.n_heads, softmax_scale=self.softmax_scale, attn_bias=attn_bias, key_padding_mask=key_padding_mask, is_causal=is_causal, dropout_p=self.attn_dropout_p, training=self.training, needs_weights=needs_weights, multiquery=True)234        return (self.out_proj(context), attn_weights, past_key_value)235 236def attn_bias_shape(attn_impl, n_heads, seq_len, alibi, prefix_lm, causal, use_sequence_id):237    if attn_impl == 'flash':238        return None239    elif attn_impl in ['torch', 'triton']:240        if alibi:241            if (prefix_lm or not causal) or use_sequence_id:242                return (1, n_heads, seq_len, seq_len)243            return (1, n_heads, 1, seq_len)244        elif prefix_lm or use_sequence_id:245            return (1, 1, seq_len, seq_len)246        return None247    else:248        raise ValueError(f'attn_impl={attn_impl!r} is an invalid setting.')249 250def build_attn_bias(attn_impl, attn_bias, n_heads, seq_len, causal=False, alibi=False, alibi_bias_max=8):251    if attn_impl == 'flash':252        return None253    elif attn_impl in ['torch', 'triton']:254        if alibi:255            (device, dtype) = (attn_bias.device, attn_bias.dtype)256            attn_bias = attn_bias.add(build_alibi_bias(n_heads, seq_len, full=not causal, alibi_bias_max=alibi_bias_max, device=device, dtype=dtype))257        return attn_bias258    else:259        raise ValueError(f'attn_impl={attn_impl!r} is an invalid setting.')260 261def gen_slopes(n_heads, alibi_bias_max=8, device=None):262    _n_heads = 2 ** math.ceil(math.log2(n_heads))263    m = torch.arange(1, _n_heads + 1, dtype=torch.float32, device=device)264    m = m.mul(alibi_bias_max / _n_heads)265    slopes = 1.0 / torch.pow(2, m)266    if _n_heads != n_heads:267        slopes = torch.concat([slopes[1::2], slopes[::2]])[:n_heads]268    return slopes.view(1, n_heads, 1, 1)269 270def build_alibi_bias(n_heads, seq_len, full=False, alibi_bias_max=8, device=None, dtype=None):271    alibi_bias = torch.arange(1 - seq_len, 1, dtype=torch.int32, device=device).view(1, 1, 1, seq_len)272    if full:273        alibi_bias = alibi_bias - torch.arange(1 - seq_len, 1, dtype=torch.int32, device=device).view(1, 1, seq_len, 1)274        alibi_bias = alibi_bias.abs().mul(-1)275    slopes = gen_slopes(n_heads, alibi_bias_max, device=device)276    alibi_bias = alibi_bias * slopes277    return alibi_bias.to(dtype=dtype)278ATTN_CLASS_REGISTRY = {'multihead_attention': MultiheadAttention, 'multiquery_attention': MultiQueryAttention}